CuPd interface charge and energy quantum entrapment: A tight-binding and XPS investigation

Applied Surface Science 257 (2010) 727–730

Contents lists available at ScienceDirect

Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc

CuPd interface charge and energy quantum entrapment: A tight-binding and XPS investigation Yanguang Nie a , Yan Wang a , Yi Sun b , Ji Sheng Pan b , B.R. Mehta c , Manika Khanuja c , S.M. Shivaprasad d , Chang Q. Sun a,e,∗ a

School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore 639798, Singapore Institute of Materials Research and Engineering, Agency for Science, Technology and Research (A*STAR), Singapore 117602, Singapore Thin Film Laboratory, Department of Physics, Indian Institute of Technology, Delhi, New Delhi 110016, India d Chemistry and Physics of Materials Unit, Jawaharlal Nehru Centre for Advanced Scientific Research, Jakkur, Bangalore 560064, India e Faculty of Materials, Optoelectronics and Physical Science, Xiangtan University, Xiangtan 411105, China b c

a r t i c l e

i n f o

Article history: Received 7 April 2010 Received in revised form 15 July 2010 Accepted 15 July 2010 Available online 22 July 2010 Keywords: Heterojunctions Nanofabrications EXAFS NEXAFS SEXAFS Electronic band structure Photoelectron spectroscopies

a b s t r a c t Materials at heterojunction interfaces demonstrate many physical and chemical properties that are indeed fascinating with mechanisms that need yet to be explored. We show herewith that the “interface charge and energy quantum entrapment due to bond order distortion and bond nature alteration” perturbs essentially the Hamiltonian and hence the binding energy of the CuPd alloy interface. Analyzing the X-ray photoelectron emission of the thermally induced evolution of the Cu 2p and Pd 3d core-level energies at the Cu–Pd interface before and after thermally alloying revealed that the Pd 3d and Cu 2p interfacial potential traps are 0.36 and 0.95 times deeper than the potential wells of the corresponding bulk constituents standing alone. © 2010 Elsevier B.V. All rights reserved.

1. Introduction Atomic impurities, interfaces of superlattices, twin grains, and embedded nanostructures can be prototyped by an atom with fully or partially coordinated heterogeneous neighbors. Many processes take place surrounding such atoms: the alloy formation associated with valence charge intermixing; compound formation with valence electron localization, repopulation, and polarization; structure distortion due to lattice mismatching, etc. Joining two materials may result in a new value of melting point (Tm ), or the cohesive energy per discrete atom, being different from that of either of the constituents standing alone [1,2]. The strength of junction interface is frequently much stronger (mechanical strength is proportional to the sum of bond energy per unit volume [3]) than either component alone, which forms the basis of superhard multilayers or stiffer nanocomposites [4,5]. Strain and chemical energy arising from the interface bond strain and bond nature alteration [6], are often used to describe the interface energy [2,7]. Interface produces locally

large perturbations to the Hamiltonian, binding energy density, and atomic cohesive energy, which determine the mechanical, thermal, electronic, optic, magnetic, dielectric, catalytic and chemical properties which are completely different from those of the constituent bulk metals [8–10]. It has been asserted that “no strain, no gain” for the FePd epitaxial interface [11], for instance. The study of surface or interface alloying is motivated by their use in many industrial applications such as catalysis, sensors, anticorrosion, friction reduction, and electronic devices. Varying composition or thermal annealing upon continuous deposition of dissimilar metals [12,13] has provided an effective method to modify interatomic strain and to redistribute charges around the bonding atoms [14,15]. Although the physics and chemistry of materials at the interfaces have been extensively investigated, the laws governing the energetic behavior of electrons and the property change of materials at the interface have not yet been established. This work aims to show that the charge and energy quantum trapping (QT) is necessary for the CuPd bimetallic interface. 2. Principle

∗ Corresponding author at: School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore 639798, Singapore. E-mail address: [email protected] (C.Q. Sun). 0169-4332/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.apsusc.2010.07.049

There has been long confusion about the origin and the reference point from which the core-level shifts upon bulk formation

728

Y. Nie et al. / Applied Surface Science 257 (2010) 727–730

or upon change of the chemical environment of the specimen. Numerous models have been developed addressing the origin as discussed in Ref. [16]. In tradition, the “initial–final” states model has been broadly used in interpreting the spectral variations. The “initial (neutral, un-ionized specimen with n electrons)–final (radiation beam ionized specimen with n − 1 electrons)” states model [17,18] defines the core-level shift as arising from the difference in cohesive energy that is needed to remove a core electron either from a surface atom or from a bulk atom. The excited surface atom is assumed as a ‘Z + 1 impurity’ sitting on the substrate meal of Z atomic number. The final states of atoms are expected to increase while the initial states to decrease when their sizes are reduced. However, the “initial–final” states model derives theoretically negative or mixed energy shift, conflicting with the trends of size dependent energy shift of nanostructures [19–24] and the trends of varying the angle between the X-ray beam and the surface normal or the incident beam energies. Intensive experimental evidence [17,18,25–32] shows that the intensity of the high-energy (smaller absolute value) component in X-ray photoelectron spectroscopy (XPS) is often enhanced with the increase of the incident photon energy or with the decrease of the angle between the incident X-ray and the surface normal. Dominated by the under-coordinated atoms, nanostructures exhibit that the core-level features move simultaneously towards lower BE when the size is reduced. These unusual observations due to atomic under-coordination seem beyond the description using the “initial–final states” effect alone. Nevertheless, the global presence of the “initial–final” states in the XPS measurement can be minimized by proper calibration of the equipment or in data processing. As discussed previously, lattice termination on flat or curved surface and the elemental mixture in alloy not only creates a potential barrier at the surface or interface but also reduces the coordination number (CN, z) of the constituent atoms [33]. The potential barrier confines electrons or holes in the surface or interface. The CN imperfection of an atom results in the remaining bonds of the lower-coordinated atom to contract spontaneously associated with magnitude increase of the binding energy [1]. The bond strength enhancement contributes not only to the atomic cohesive energy (Ecoh , single bond energy multiplies the atomic CN) of the specific atom but also to the energy density in the relaxed surface region. The binding energy density perturbs the Hamiltonian of an extended solid that determines the entire band structure such as the band gap, core-level shift, and bandwidth [34,35]. Following the tight-binding approximation [36], the core-level shift is determined uniquely by the exchange and overlap integrals (see Eq. (1)) that involve the eigen wave function,  (ri ), and the interatomic potential, Vcry (r). All the energy shift proceeds from the binding energy of an isolated atom, E (0), due to intraatomic trapping, rather than from that of the bulk E (B). The energy shift happens upon a bulk formation because the involvement of interatomic interaction, and hence, / 0. E (B) = E (B) − E (0) = Fig. 1a illustrates the proposed interface QT mechanism. In the interface region, there is a gradient of chemical composition due to the diffusion of the constituent atoms. E0 is the vacuum level; Vatom (r) is the potential of intraatomic trapping. Including the many-body interaction, Vcry (r, B) represents the periodic interatomic potential experienced by electrons in the constituent bulk (B). Moving cross the interface region from one to the other constituent, the Vcry (r, B) evolves into Vcry (r, I) at the middle of the interface (I) because of valence charge intermixing or bond nature alteration. The coefficient  shown in Fig. 1a represents the interface bond-energy ratio to that of the constituent bulk. The Vcry (r, I) may be deeper ( > 1) or shallower ( < 1) than the corresponding Vcry (r, B) for a specific constituent. If dipole is produced in the

Fig. 1. Schematic illustration of (a) the QT model and (b) the QT-induced core-level shift, E (I) and E (B) from the energy level of an isolated atom, E (1), Vatom (r) is the intraatomic potential. Vcry (r, B) and Vcry (r, I) are interatomic potentials in the bulk and the interface region, respectively.  is the enhance parameter.  > 1 represents the compact and  < 1 represents the dissociated interface. Annealing will increases the E (I) intensity, rendering that of the E (B), upon alloy formation.

interface region [37],  < 1; otherwise,  > 1. In the present study, we will focus on the case of  > 1. According to the tight-binding theory [36], the shift of a particular th level, E (B), from the binding energy of an isolated atom, E (0), as indicated in Fig. 1b, is determined by the exchange and overlap integrals [16],

⎧  h¯   2 ⎪ − + V (r) + Vcry (r)(1 + H )  H = ⎪ atom ⎪ 2m ⎪ ⎪ ⎪ ⎪ ⎨ E (0) =  (r )|Vatom (r)|v (r ) i

i

⎪ E (I) = E (B) = (ˇ + z˛)

(core -level shift)

⎪ ⎪ ⎪ ⎪ ˇ = − (ri )|Vcry (r)[1 + H ] (ri ) > 0 (overlap integral) ⎪ ⎪ ⎩

˛ = − (ri )|Vcry (r)[1 + H ] (ri ) > 0 (exchange integral) (1)

where  (ri ) is the specific Bloch wave function at site ri , which satisfies,  (ri )| (rj )  = ıij (if i = j, ıij = 1; otherwise, ıij = 0) with   being the charge density. H =  − 1 is the Hamiltonian (H) perturbation, arising from the interface bond strain and bond nature alteration [38]. In the interface region, both the atomic coordination number z and the Bloch wave function are changed insignificantly, the binding energy shift is thus uniquely determined by the perturbed potential-well depth or the bond energy, E (I) =  = 1 + H E (B)

or

E (I) − E (B) = H =  − 1 E (B)

(2)

Based on this expression, we can determine the interface bond strength based on the XPS decomposition of the E (B) and the E (I) − E (B), as illustrated in Fig. 1b. It is expected that the intensities for the peaks of E (I) and E (B) evolve upon annealing; the intensity of E (I) peak will increase rendering that of E (B) as

Y. Nie et al. / Applied Surface Science 257 (2010) 727–730

729

Table 1 XPS measured intensity (I) evolution of Pd 3d5/2 and Cu 2p3/2 binding energy (BE) at the Cu(2 nm)–Pd(10 nm) interface as a function of annealing temperature, T. T (K)

340 540 573 673 940 Mean

Pd 3d5/2 (eV)

Cu 2p3/2 (eV)

Bulk

Interface

335.67 335.66 335.63 335.58 335.58 335.62

337.10 337.17 337.19 337.22 337.26 337.19

IB /II 15.30 5.06 4.78 1.56 0.18 –

Bulk

Interface

931.65 931.57 931.62 931.60 931.65 931.61

933.20 933.21 933.30 933.30 933.19 933.23

IB /II 11.50 2.93 3.23 1.40 0.282 –

the total number of electrons in the particular band conserves. The intensity evolution reflects the extent of alloy formation. 3. Experimental In order to verify the proposed mechanism of interface QT, Cu(2 nm)/Pd(10 nm) thin films were firstly deposited onto Si wafer using a physical–vapor–deposition system [13]. XPS spectra were collected then as a function of isochronal annealing (1 min) at increasingly high temperature up to 940 K (with XPS spectra being recorded on cooling to 300 K). Each spectrum was decomposed into two Gaussian components, E (I) and E (B), as summarized in Table 1. At room temperature, no significant fingerprints of alloy formation (peak I) were observed but with the increase of annealing temperature, additional peaks emerge at energies of 1.62 eV below the Cu 2p3/2 and 1.57 eV below that of Pd 3d5/2 . As the annealing temperature keeps on increasing, intensities of the E (I) peak increase rendering that of E (B). 4. Results and discussion Fig. 2 shows the typical Cu 2p3/2 and Pd 3d5/2 spectra collected after 540 K and 940 K annealing. The intensity reversions of the E (I) and the E (B) at 540 K and 940 K indicate the completeness of interface alloying. In addition, low-energy electron and photoelectron diffraction studies [39,40] revealed that the Cu–Pd interlayer distance contracts up to 7 ± 2.5%. The interface alloy is formed through atomic interdiffusion [41] with an estimated activation energy of 0.88 eV [42,43]. A theoretical reproduction [16] of the measured size dependence of Cu 2p3/2 [44] and Pd 3d5/2 surface [45] core-level shift has derived that the E (B) for Cu 2p3/2 and Pd 3d5/2 are 1.70 eV and 4.36 eV. These data allows us to estimate the interface potentialwell depth for the Pd 3d and Cu 2p electrons with the resolved E (I) − E (B) as given in Table 1. Calculation results in Table 2 indicate that the  values for Pd and Cu are 1.36 and 1.95, respectively. The corresponding core-level shift are 5.93 eV and 3.32 eV in comparison to the bulk shift of 4.36 eV and 1.70 eV for the Pd 3d and the Cu 2p levels in the interface region. Apparently, such a huge energy shift due to alloy formation is beyond the description of the well-received “initial–final states” or the “hole screening” mechanism [46] that comes into play throughout the measurement and hardly sheds the alloying effect. The interface QT mechanism agrees with the proposal of a wedge-shaped potential well for the monolayer structure sandwiched between the SrTiO3 and the LaTiO3 superlattices in Table 2 Summary of the relative depths  and the interface energy shift E(I) for the Pd 3d5/2 and Cu 2p3/2 core levels. E-Level (eV)

E (B)

E (I)

E (I) − E (B)



E (I)

Pd 3d5/2 [45] Cu 2p3/2 [16,44]

4.36 1.70

330.34 931.00

1.57 1.62

1.36 1.95

5.93 3.32

Fig. 2. Thermally driven spectral evolution of (a) Cu 2p3/2 and (b) Pd 3d5/2 BE upon annealing at 540 K and 940 K and their decompositions.

calculations [47]. The QT was suggested to arise from the Coulomb potential of a two-dimensional charged La sheet, which in turn confines the electrons in the Airy-function-localized states. Here we suggest that the interface QT arises from the bond strain [39,40] and valence charge intermixing. As a consequence of the QT, energy levels of atoms in the interface region will go deeper or shift positively, accordingly. On the other hand, the interfacial atomic cohesive energy, EC,I = zEb , will increases as the cohesive energy per bond increases from the bulk value of Eb to EI = Eb and the z changes insignificantly [1]. The EC,I determines the Tm uniquely. Therefore, the interface QT can be further verified by the observed size dependence of overheating of embedded nanostructures. A theoretical analysis [38] of the size-induced overheating for Ag/Ni, and Pb/Al and Pb/Zn coreshell nanostructures has led to a  = 1.8 for the Ag and Pb cores, indicating that an interfacial bond is 80% stronger than a bond in the bulk of the core material, which is in line with the current derivation of  > 1 for the CuPd interface. 5. Summary We have thus demonstrated that the interface charge and energy QT is essential for the Cu–Pd interface and that the trap depth is, respectively, 1.36 and 1.95 times that of the bulk Pd and Cu. The QT arises from the interface bond order distortion and valence charge intermixing that perturbs the Hamiltonian, the atomic cohe-

730

Y. Nie et al. / Applied Surface Science 257 (2010) 727–730

sive energy, and hence the related properties. This approach not only provides a feasible way of interface bond strength determination but also a consistent insight into the energetic behavior of electrons in the interface region. This straight forward yet simple method can be used to extract information regarding the local energy and charge of other interfaces or bimetallic alloys. Acknowledgments Financial supports from Nanyang Technological University (RG16/09) and A*Star, Singapore, Nature Science Foundation (No. 10772157) of China, are all gratefully acknowledged. References [1] C.Q. Sun, Progress in Solid State Chemistry 35 (1) (2007) 1–159. [2] Q. Jiang, H.M. Lu, Surface Science Reports 63 (10) (2008) 427–464. [3] X.J. Liu, J.W. Li, Z.F. Zhou, L.W. Yang, Z.S. Ma, G.F. Xie, Y. Pan, C.Q. Sun, Applied Physics Letters 94 (2009), 131902(13). [4] S. Veprek, A.S. Argon, Journal of Vacuum Science and Technology B 20 (2) (2002) 650–664. [5] L. Ci, J. Suhr, V. Pushparaj, X. Zhang, P.M. Ajayan, Nano Letters 8 (9) (2008) 2762–2766. [6] C.Q. Sun, Progress in Materials Science 54 (2) (2009) 179–307. [7] G. Ouyang, X. Tan, C.X. Wang, G.W. Yang, Chemical Physics Letters 420 (1–3) (2006) 65–70. [8] J.G. Chen, C.A. Menning, M.B. Zellner, Surface Science Reports 63 (5) (2008) 201–254. [9] J.A. Rodriguez, D.W. Goodman, Science 257 (5072) (1992) 897–903. [10] P. Kamakoti, B.D. Morreale, M.V. Ciocco, B.H. Howard, R.P. Killmeyer, A.V. Cugini, D.S. Sholl, Science 307 (5709) (2005) 569–573. [11] J. Buschbeck, I. Opahle, M. Richter, U.K. Rößler, P. Klaer, M. Kallmayer, H.J. Elmers, G. Jakob, L. Schultz, S. Fähler, Physical Review Letters 103 (2009) 216101 (Copyright (C) 2010 The American Physical Society). [12] G. Liu, T.P. St Clair, D.W. Goodman, Journal of Physical Chemistry B 103 (40) (1999) 8578–8582. [13] M. Khanuja, B.R. Mehta, S.M. Shivaprasad, Thin Solid Films 516 (2008) 5435–5439. [14] D.H. Zhang, W. Shi, Applied Physics Letters 73 (8) (1998) 1095–1097. [15] Y.X. Dang, W.J. Fan, S.T. Ng, S.F. Yoon, D.H. Zhang, Journal of Applied Physics 97 (10) (2005) 103718. [16] C.Q. Sun, Physical Review B 69 (2004), 045105(4). [17] M. Alden, H.L. Skriver, B. Johansson, Physical Review Letters 71 (15) (1993) 2449–2452. [18] E. Navas, K. Starke, C. Laubschat, E. Weschke, G. Kaindl, Physical Review B 48 (19) (1993) 14753–14755. [19] B. Balamurugan, T. Maruyama, Journal of Applied Physics 102 (2007), 034306(3).

[20] C. Bittencourt, A. Felten, B. Douhard, J. Ghijsen, R.L. Johnson, W. Drube, J.J. Pireaux, Chemical Physics 328 (1–3) (2006) 385–391. [21] S.P. Suprun, E.V. Fedosenko, Semiconductors 41 (5) (2007) 590–595. [22] I. Aruna, B.R. Mehta, L.K. Malhotra, S.M. Shivaprasad, Journal of Applied Physics 104 (2008) (6, 064308). [23] J.G. Tao, J.S. Pan, C.H.A. Huan, Z. Zhang, Y. Sun, J.W. Chai, S.J. Wang, Journal of Physics-Condensed Matter 20 (48) (2008). [24] J.G. Tao, J.S. Pan, C.H.A. Huan, Z. Zhang, J.W. Chai, S.J. Wang, Surface Science 602 (16) (2008) 2769–2773. [25] B.S. Fang, W.S. Lo, T.S. Chien, T.C. Leung, C.Y. Lue, C.T. Chan, K.M. Ho, Physical Review B 50 (15) (1994) 11093–11101. [26] B. Balamurugan, T. Maruyama, Applied Physics Letters 89 (2006), 033112(3). [27] R.A. Bartynski, D. Heskett, K. Garrison, G. Watson, D.M. Zehner, W.N. Mei, S.Y. Tong, X. Pan, Journal of Vacuum Science and Technology A 7 (3) (1989) 1931–1936. [28] J.N. Andersen, D. Hennig, E. Lundgren, M. Methfessel, R. Nyholm, M. Scheffler, Physical Review B 50 (23) (1994) 17525–17533. [29] D.M. Riffe, G.K. Wertheim, Physical Review B 47 (11) (1993) 6672–6679. [30] J.H. Cho, K.S. Kim, S.H. Lee, M.H. Kang, Z.Y. Zhang, Physical Review B 61 (15) (2000) 9975–9978. [31] A.V. Fedorov, E. Arenholz, K. Starke, E. Navas, L. Baumgarten, C. Laubschat, G. Kaindl, Physical Review Letters 73 (4) (1994) 601–604. [32] L.I. Johansson, H.I.P. Johansson, J.N. Andersen, E. Lundgren, R. Nyholm, Physical Review Letters 71 (15) (1993) 2453–2456. [33] C.Q. Sun, Physical Review B 69 (2004) 045105 (Copyright (C) 2010 The American Physical Society). [34] L. Pauling, Journal of the American Chemical Society 69 (3) (1947) 542–553. [35] V.M. Goldschmidt, C Berichte Der Deutschen Chemischen Gesellschaft 60 (1927) 1263–1296. [36] M.A. Omar, Elementary Solid State Physics: Principles and Applications, Addison-Wesley, New York, 1993. [37] C.Q. Sun, Y. Wang, Y.G. Nie, B.R. Mehta, M. Khanuja, S.M. Shivaprasad, Y. Sun, J.S. Pan, L.K. Pan, Z. Sun, Physical Chemistry Chemical Physics 12 (13) (2010) 3131–3135. [38] C.Q. Sun, Y. Shi, C.M. Li, S. Li, T.C.A. Yeung, Physical Review B 73 (2006), 075408(7). [39] A. de Siervo, E.A. Soares, R. Landers, G.G. Kleiman, Physical Review B 71 (2005), 115417(11). [40] A. de Siervo, R. Paniago, E.A. Soares, H.D. Pfannes, R. Landers, G.G. Kleiman, Surface Science 575 (1–2) (2005) 217–222. [41] J.P. Reilly, C.J. Barnes, N.J. Price, R.A. Bennett, S. Poulston, P. Stone, M. Bowker, Journal of Physical Chemistry B 103 (31) (1999) 6521–6532. [42] M.L. Grant, B.S. Swartzentruber, N.C. Bartelt, J.B. Hannon, Physical Review Letters 86 (20) (2001) 4588–4591. [43] S.V. Eremeev, G.G. Rusina, E.V. Chulkov, Surface Science 601 (17) (2007) 3640–3644. [44] D.Q. Yang, E. Sacher, Applied Surface Science 195 (1–4) (2002) 187–195. [45] C.Q. Sun, Y. Wang, Y.G. Nie, Y. Sun, J.S. Pan, L.K. Pan, Z. Sun, Journal of Physical Chemistry C 113 (52) (2009) 21889–21894. [46] M.V. Ganduglia-Pirovano, J. Kudrnovsk, M. Scheffler, Physical Review Letters 78 (9) (1997) 1807. [47] Z.S. Popovic, S. Satpathy, Physical Review Letters 94 (2005), 176805(17).