Acta Materialia 52 (2004) 501–505 www.actamat-journals.com Size-effect on the electronic structure and the thermal stability of a gold nanosolid Chang Q. Sun a a,* , H.L. Bai b, S. Li a, B.K. Tay a, E.Y. Jiang b School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore 639798, Singapore b Institute of Advanced Materials Physics and Faculty of Science, Tianjin University, Tianjin 30072, PR China Received 16 July 2003; received in revised form 17 September 2003; accepted 19 September 2003 Abstract Consistent insight into the size-enhanced E4f -level shift and the size-suppressed melting point of Au nanosolids has been obtained based on the bond order-length-strength (BOLS) correlation mechanism [Sun et al., Acta Mater. 51 (2003) 4631]. Consistency between theory predictions and observations reveals that the atomic-coordination number (CN)-imperfection induced bond contraction and the associated bond energy increase dictate these changes. The increase of binding energy density per unit volume in the relaxed surface region perturbs the Hamiltonian that determines the core-level shift; the decrease of atomic cohesive energy (a product of atomic CN and the single bond energy) determines the thermal energy required for melting. Extending the knowledge to the lower end of the size limit suggests that the metallic bond in the gold monatomic-chain contracts by 30% associated with 43% magnitude rise of the bond energy. The crystal binding intensity contributing from all the atoms in the solid to the Au-E4f electrons in the bulk is determined to be )2.87 eV. Ó 2003 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: Nanostructure; Electronic structure; Thermal stability; Mechanical strength Metallic nanosolids, or nanoclusters, nanocrystals or nanoparticles, have attracted tremendous interest, since they show intriguing properties from a basic scientific viewpoint, as well as great potential in upcoming technological applications such as nanoelectronics. The intriguing phenomena include quantum conductance, enhanced mechanical strength and binding energy, and the lowered chemical reactivity and thermal stability. Quantifying the intensity of crystal binding energy to a core electron of an isolated atom and the core-level energy shift upon bulk and nanosolid formation are of great challenge. Although several models have been developed for the size dependence of melting-point (Tm ) suppression [1,2] and core-level shift, [3] consistent understanding of the electronic structure, mechanical strength, and thermal stability of a metallic nanosolid is highly desirable. In this presentation, we describe an * Corresponding author. Tel.: +65-6790-4517; fax: +65-6792-0415. E-mail address: [email protected] (C.Q. Sun). URL: http://www.ntu.edu.sg/home/ecqsun. effective approach to determine the binding energy of a core electron and the thermal stability of gold nanosolids by decoding the measured size dependence of both the Au 4f-level energy [4] and the melting-point suppression based on the recent bond order-length-strength (BOLS) correlation [5–7]. Decoding the size dependence of the 4f core-level shift gives information about the 4flevel energy of an Au atom isolated from the solid and the crystal binding intensity of the entire solid to the 4f electron upon bulk formation. Extrapolation of the size dependent Tm suppression suggests that the gold monatomic chain (MC) melt at a temperature of 320 K. The BOLS correlation that is derived from the correlation of atomic-coordination number (CN)–atomicradius noted by Goldschmidt [8] and Febelman [9] indicates that the CN-imperfection of atoms at the surface or at sites surrounding defects causes the remaining bonds of the lower-coordinated atom to contract to di ¼ ci d, spontaneously. The spontaneous bond contraction is associated with magnitude increase of the bond energy, or potential well deepening, ei ¼ cm i eb . 1359-6454/$30.00 Ó 2003 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.actamat.2003.09.033 502 C.Q. Sun et al. / Acta Materialia 52 (2004) 501–505 The m is an adjustable parameter depending on the nature of the bond of the system. For elemental metals, m ¼ 1. According to Goldschmidt, if an atom reduces its CN from 12 to 8, 6, and 4, the radius of the specific atom will contract by 3%, 4%, and 12%, respectively. The dimmer bond of metals such as Ti, Zr and V has been reported to contract by 30–40% [9]. The BOLS correlation is represented by ( ci ðzi Þ ¼ di =d0 ¼ 2=½1 þ expðð12 zi Þ=8zi Þ; ð1Þ ei ¼ cm i eb : The i denotes the atom in the ith atomic layer. The ei and eb are the zero-temperature binding energies per bond which is independent of the type of the pair-wise interatomic potential. d0 is the corresponding bulk value of bond length. For a flat or a slightly curved surface, z1 ¼ 4 [1,5,10,11]; for a spherical dot, z1 ¼ 4ð1 0:75=Kj Þ [12], z2 ¼ 6 and z3 ¼ 12. Kj ¼ Rj =d0 is the number of atoms lined along the radius of a spherical dot. The striking difference between a bulk solid and a spherical nanosolid is that the portion of the lowercoordinated atoms is considerably large for a nanosolid. Generally, the mean relative change of a detectable quantity Q of a nanosolid containing Nj atoms can be expressed as QðRj Þ ¼ Nj q, and as Qð1Þ ¼ Nj q, for the same solid without considering the effect of surface CN-imperfection. Using a shell structure, the QðRj Þ is expressed as X Nj q ¼ Nj q þ Ni ðqi qÞ; ð2Þ i63 the relative change of the QðRj Þ follows a scaling law [5]: 8 P DQðRj Þ > < Qð1Þ ¼ i 6 3 cij ðqi =q 1Þ ¼ Dqj ; ð3Þ 1 ðKj 6 3Þ; Rsout;i Rsin;i Ni > c ¼ ¼ : ij Nj i Rsj / sc ðK > 3Þ; j Kj where q and qi correspond to the density of the Q inside the bulk and in the ith surface layer, respectively. s ¼ 1; 2 and 3 correspond to dimensionality of a thin plate, a rod, and a spherical dot. The i is counted up to three from the outermost atomic layer to the center of the solid as no CNimperfection is expected when i > 3. Rin;i and Rout;i correspond to the inner and outer radius of the ith atomic layer (Rout;i Rin;i ¼ di ). Eq. (3) represents that the sizeand-shape dependence of a detectable quantity of a nanosolid originates from the Dqi =q that determines the sigh and magnitude of change. Such a change follows the portion of surface atoms, cij , of the entire solid. For a spherical dot at the lower end of the size limit, Kj ¼ 1:5 (Rj ¼ Kj d0 ¼ 0:43 nm for an Au spherical dot), c1j ¼ 1, c2j ¼ c3j ¼ 0, and z1 ¼ 2, which is identical in situation to an atom in a monatomic chain despite the geometrical orientation of the two interatomic bonds. Actually, the geometrical orientation of the bond configuration contributes not to the modeling exercise. Therefore, the performance of an atom in the smallest nanosolid is a mimic of an atom in a monatomic chain of the same element without presence of external stimulus such as stretching or heating. As a scaling law, the measured size dependence of the QðRj Þ is always proportional to the inverse Rj (converges at Kj < 3) with a slope b. Combining the scaling law based on measurement and theory of Eq. (3), we have 1 bRj ðmeasurementÞ; QðRj Þ Qð1Þ ¼ ð4Þ Qð1Þ Dqj ðtheoryÞ with Qð1Þ b=ðDqj Rj Þ. The Dqj / R1 j (Eq. (3)) varies simply with the parameter m and the known dimensionality and known size of the solid. There are only two independent variables, m and Qð1Þ. If a certain known Qð1Þ value such as the Tm ð1Þ of the considered system is given, the m can be readily obtained. With the obtained m, any other unknown quantities Qð1Þ such as the crystal binding intensity, E4f ð1Þ, of the same system can be determined uniquely with the above relations. Not surprisingly [5], the size-and-shape dependence of a detectable quantity of the same system can also be predicted for materials design purpose once the corresponding qðz; eÞ function is established. The CN-imperfection and the associated bond energy rise contributes not only to the cohesive energy (Ecoh;i ¼ zi ei ) per atom but also to the binding energy density (EB;i ¼ ni ei ) per unit volume in the relaxed region. The atomic Ecoh;i determines the thermodynamic behavior of a nanosolid such as melting, or phase transition (Tm / Ecoh ) [1]. The compressibility (under pressing force) or extensibility (under stretching force) at constant temperature is given by 1 1 oV o2 u ds b¼ ; ¼ V / V oP T oV 2 e T ou P¼ ; oV T ð5Þ b is the inverse YoungÕs modulus that increases at surfaces due to the CN-imperfection enhanced bond energy [5,13]. The EB;i contributes to the overall potential in the Hamiltonian of an extended solid which leads to the perturbation of the core-level energy of the solid [10] DEm ðRj Þ ¼ DEm ð1Þb1 þ DHj c; ð6Þ where DEm ð1Þ ¼ Em ð1Þ Em ð1Þ is the energy shift upon bulk formation. The DHj , being independent of the particular form of the interatomic potential, is the contribution from bond relaxation of the lower-coordinated atoms at the curved surface of the nanosolid. As such, the bond contraction, Ddj , the melting-point suppression, DTmj , the Hamiltonian perturbation, DHj , and the compressibility/extensibility modification, Dbj , of a nanosolid are in the form: C.Q. Sun et al. / Acta Materialia 52 (2004) 501–505 8 P di d0 P > Ddj ¼ cij d0 ¼ cij ðci 1Þ < 0; > > > i63 i63 > > P ni ei nb e0 P > > > DHj ¼ cij nb e0 ¼ cij ðnib cm 1Þ > 0; > i < i63 i63 P zi ei zb e0 P > cij zb e0 ¼ cij ðzib cm 1Þ < 0; DTmj ¼ i > > i 6 3 i 6 3 > > > > P dis e1 P > d s e1 > > cij id s e10 0 ¼ cij ðcsþm 1Þ < 0: : Dbj ¼ i i63 0 0 ð7Þ i63 One needs to note that the bond number density in the relaxed region does not change upon relaxation and hence ni =nb ¼ 1. Incorporating the above relations into the measured size-and-shape dependence of the corresponding properties, we can obtain quantitative information that is beyond direct measurement. As discussed, at the lower end of the size limit (z ¼ 2, c1j ¼ 1 and c2j ¼ c2j ¼ 0), all the property changes are related to the behavior of a single bond of the lower-coordinated atoms. For instance, with the measured size dependence of the core-level shift, we can discriminate the crystal binding (bulk shift) from atomic trapping (core-level position of an isolated atom) of an isolated atom [10]. We may let Qð1Þ ¼ DE4f ð1Þ ¼ E4f ð1Þ E4f ð1Þ in Eq. (4) for a Au solid. Least-mean-square linearization of the measured E4f ð1=Rj Þ of Au nanosolids deposited on various substrates such as octanedithiol [4], TiO2 [14], Pt [15], and thiol-caped Au particles [16] gives intercepts a that correspond to the bulk value of E4f ð1Þ ¼ 84:37 eV [17], and a slopes b that vary with the dimensionality of the Au solids grown on different substrates. With the known m ¼ 1 value for pure metals [5], the DE4f ð1Þ and E4f ð1Þ can be determined as given in Table 1. Comparing the two theoretical curves for a spherical dot and a thin plate with the measured profiles in Fig. 1(a) reveals that the Au solid on octanedithiol substrate follows ideally the curve of a spherical dot. The BOLS correlation in Eq. (1) suggests that at R ¼ 1:5d ¼ 0:43 nm, the Au–Au distance contracts by 30% from 0.288 to 0.200 nm, which is closes to the value, 0.23 0.04 nm, measured at 4.2 K [18] and the calculated shorter distance of 0.232 nm [19] as well. Table 1 The length and energy of the Au–Au bond in the monatomic chain and the core-level energy of an isolated Au atom obtained from decoding the E4f ðRj Þ and Tm ðRj Þ of nanosolid Au b m s E4f (eV) DE4f ð1Þ (eV) Dchain (nm) echain =ebulk Tm;chain =Tm ð1Þ 503 Au/ octanedithiol Au/TiO2 3.7804 1 3 )81.504 )2.866 0.2001 1.43 1/4.2 1.5253 1 )81.506 )2.864 Au/ Pt Au/ thiol 1 )81.504 )2.866 3 )81.505 )2.865 Fig. 1. Comparison of theory with the measured size dependence of (a) ½E4f ðKÞ E4f ð1Þ=½E4f ð1Þ E4f ð1Þ of Au (nanodot) on octanedithiol [4] and Au (nanoplate) on TiO2 [14] and Pt [15] substrates and thiolcaped [16] with derived information as given in Table 1, and (b) [Tm ðKÞ Tm ð1Þ=Tm ð1Þ of Au nanosolids on W [21] and C [22] substrates and embedded in SiO2 matrix [23], showing strongly interfacial effects and dimensionality transition on the Au nanosolid melting. K ¼ R=d: Melting at the lower end of size limit (K ¼ 1:5; z ¼ 2) corresponds to the situation of a gold monatomic chain. The cohesive energy per bond increases by 43%, which coincides with the measured E4f from the smallest Au nanosolid as shown in Fig. 1(a). The deviation between theory and measurement from the thinner Au/TiO2 film is due to the initial island-like growth mode of metal on oxide surface [14,20], as well as the Au/TiO2 interfacial effect. Measurements in Fig. 1(a) show the interesting trend that the core-level drops abruptly from that of an isolated atom by a maximum (40%) upon the smallest nanosolid being formed and then the shift recovers in a R1 fashion to the bulk value when the solid grows from atomic scale to macroscopic size. Calibrated with Qð1Þ ¼ Tm ð1Þ ¼ 1337:33 K and using the same mð¼ 1Þ value used in calculating the E4f (Rj ), we obtained the theoretical Tm -suppression curves for different shapes, which are compared in Fig. 1(b) with the measured size-dependent Tm of Au on W [21] and on C [22] substrates, and Au encapsulated in silica matrix [23]. Differing from the core-level shift data, the melting profiles show that at the smaller size, the Au/W interface promotes more significantly the melting of Au (super-cooling) than the Au/C interface. The silica matrix causes slightly super-heating of the embedded Au solid compared with the rest two substrates. Besides, the Tm measurement is a thermodynamic process that may 504 C.Q. Sun et al. / Acta Materialia 52 (2004) 501–505 affect the results to some extent. Nevertheless, the theoretically predicted melting curves merge at the lower end of the size limit, K ¼ 1:5 with 75% suppression. Therefore, it is anticipated that thermal rupture of the Au–Au chain occurs at 320 K (to Tmb zib c1 i ¼ 2Tmb =ð12 0:7Þ ¼ Tmb =4:2), much lower than the gold bulk melting point (1337.33 K). According to Eq. (7), the compressibility/extensibility of the monatomic chain is 0.5 ( ¼ 0:72 1) times the bulk values (for a metallic monatomic chain, s ¼ 1, m ¼ 1). The shortened bond is twice stronger 2 (strength ¼ ei =di / c2 2), agreeing with prei ¼ 0:7 dictions of [24]. This means that more force is required for stretching or compressing by the same length a single bond in the atomic chain compared to the force needed to stretch the same single bond in the bulk by the same amount, in general. Whereas, high-pressure X-ray diffraction revealed that the lattice constant of small alumina solid is easier to compress than larger ones [25], which disagrees with the current prediction. However, it should be noted that the temperature (Tsm ) of a solid in the semi-solid state is always lower than the Tm that is suppressed by reducing particle size. At Tsm , the interatomic bond can be stretched and unfolded more easily, and hence, the plasticity or the plastic energy of a suspended atomic chain is much higher than the atomic chain embedded in the solid. Meanwhile, the heat released from bond unfolding and stretching should raise the actual temperature of the specimen as well. Therefore, relation (Eq. (4)) for the constant temperature compressibility or extensibility no longer holds for atomic chains or nanowires of which the Tsm < Tm is suppressed by size reduction. In reality, the breaking limit of a bond is determined by Z dp ðT Þ Eplastic ðT Þ ¼ f ðxÞ dx; ð8Þ dk ðT Þ where dk ðT Þ and dp ðT Þ are the corresponding elastic and plastic limit at T. The Eplastic remains constant at a given temperature, while dp ðT Þ depends on both f ðxÞ and T . Any fluctuation of the T or the stretching force f ðxÞ applied in measurement will affect the measured dp ðT Þ, and therefore, it understandable now why the roomtemperature Au–Au breaking limit (vary from 0.29 nm [26], 0.36 nm (30%) [27], 0.35–0.40 nm [28] to even a single event of 0.48 nm [29]) is much longer than that at 4 K (0.23 0.4 nm). In the current BOLS approach, we assumed that contribution from particle–substrate interaction is negligible. Actually, the interfacial reaction modifies the m value slightly [3]. Matching predictions to measurements in Figs. 1(a) and (b) evidence the validity of the assumption. Taking the temperature effect on the total energy of a single bond into account would be significant in understanding the high ductility and the exten- sibility/compressibility of metallic nanowires. Further investigation is in progress. We have thus incorporate the effect of atomic CNimperfection to the core-level shift and the melting point of gold nanosolids, which derives a new method calibrating the DE4f ð1Þ and the E4f ð1Þ of an Au atom isolated from the solid with a prediction of the length, the strength, the extensibility and the thermal stability of the Au–Au single bond under the conditions with and without external stimuli, by simultaneously decoding the known size dependence of the E4f ðRj Þ and Tm ðRj Þ. Findings provide consistent insight into the CN-imperfection-enhanced binding intensity, mechanical strength, the suppressed thermal stability, and the compressibility/extensibility of gold nanosolids, which could be extended to the thermal and mechanical behavior of other metallic nanowires. Practice should be a helpful exercise for information of bonding identities and the single electron energy of an isolated atom that determine the performance of nanosolid materials. References [1] Sun CQ, Wang Y, Tay BK, Li S, Huang H, Zhang YB. J Phys Chem B 2002;106:10701, references there in. [2] Jiang Q, Liang LH, Zhao M. J Phys: Condens Matter 2001;13:L397. [3] Sun CQ, Pan LK, Bai HL, Li ZQ, Wu P, Jiang EY. Acta Mater 2003;51:4631, references there in. [4] Ohgi T, Fujita D. Phys Rev B 2002;66:115410. [5] Sun CQ. Prog Mater Sci 2003;48:521–685. [6] Sun CQ, Tay BK, Zeng X, Li S, Chen TP, Zhou J, Bai H, Jiang EY. J Phys: Condens Matter 2002;14:7781. [7] Sun CQ, Li S, Tay BK, Chen TP. Acta Mater 2002;50:4687. [8] Goldschmidt VM. Ber Deut Chem Ges 1927;60:1270. [9] Feibelman PJ. Phys Rev B 1996;53:13740. [10] Sun CQ, Tay BK, Fu YQ, Li S, Chen TP, Bai HL, Jiang EY. J Phys Chem B 2003;107:411. [11] Pan LK, Sun CQ, Tay BK, Chen TP, Li S. J Phys Chem B 2002;106:11725. [12] Sun CQ, Pan LK, Fu YQ, Tay BK, Li S. J Phys Chem B 2003;107:5113. [13] Sun CQ, Tay BK, Lau SP, Sun XW, Zeng XT, Bai H, Liu H, Liu ZH, Jiang EY. J Appl Phys 2001;90:2615. [14] Howard A, Clark DNS, Mitchell CEJ, Egdell RG, Dhanak VR. Surf Sci 2002;518:210. [15] Salmon M, Ferrer S, Jazzar M, Somojai GA. Phys Rev B 1983;28:1158. [16] Zhang P, Sham TK. Phys Rev Lett 2003;90:245502. [17] Doniach S, Sunjic M. J Phys 1970;4C31:285. [18] Untiedt C, Yanson AI, Grande R, Rubio-Bollinger G, Agra€it N, Vieira S, van Ruitenbeek JM. Phys Rev B 2002;66:085418. [19] Bahn S, Jacobsen KW. Phys Rev Lett 2001;87:266101. [20] Becker C, Rosenhahn A, Wiltner A, von Bergmann K, Schneider J, Pervan P, Milun M, Kralj M, Wandelt K. New J Phys 2002;4:75. [21] Castro T, Reifenberguer R, Choi E, Andres RP. Phys Rev B 1990;42:8548. [22] Buffat Ph, Borel JP. Phys Rev A 1976;13:2287. [23] Dick K, Dhanasekaran T, Zhang Z, Meisel D. J Am Chem Soc 2002;124:2312. C.Q. Sun et al. / Acta Materialia 52 (2004) 501–505 [24] Rubio-Bollinger G, Bahn SR, Agra€it N, Jacobsen KW, Vieira S. Phys Rev Lett 2001;87:026101. [25] Chen B, Penwell D, Benedetti LR, Jeanloz R, Kruger MB. Phys Rev B 2002;66:144101. [26] Takai Y, Kadwasaki T, Kimura Y, Ikuta T, Shimizu R. Phys Rev Lett 2001;87:106105. 505 [27] Yanson AI, Rubio-Bollinger G, van den Brom HE, Agra€it N, van Ruitenbeek JM. Nature (London) 1998;395:783. [28] Ohnishi H, Kondo Y, Takayanagi K. Nature (London) 1998;395:780. [29] Legoas SB, Galvao DG, Rodrigues V, Ugarte D. Phys Rev Lett 2002;88:076105.
© Copyright 2026 Paperzz