J. Phys. Chem. A 2010, 114, 8417–8422
8417
Magnetic Properties of CumOn Clusters: A First Principles Study Fan Yang,† Qiang Sun,‡ L. L. Ma,† Yu Jia,‡ S. J. Luo,§ J. M. Liu,§ W. T. Geng,| J. Y. Chen,⊥ Sa Li,†,# and Ying Yu*,† Institute of Nanoscience and Nanotechnology, Huazhong Normal UniVersity, Wuhan 430079, China, School of Physics and Engineering, Zhengzhou UniVersity, Zhengzhou 450001, China, Laboratory of Solid State Microstructures, Nanjing UniVersity, Nanjing 210093, China, Materials Modeling Laboratory, UniVersity of Science and Technology Beijing, Beijing 100083, China, School of EnVironment and CiVil Engineering, Wuhan Institute of Technology, Wuhan 430073, China, and Department of Physics, Virginia Commonwealth UniVersity, Richmond, Virginia 23284 ReceiVed: April 24, 2010; ReVised Manuscript ReceiVed: July 8, 2010
Experimental evidence shows that small Cu2O nanoparticles exhibit ferromagnetic or paramagnetic properties, allowing for the promising possibility to recycle the catalyst Cu2O easily in wastewater treatment. In this paper, theoretical calculation studying the magnetic property of copper/oxide clusters is reported. A series of CumOn ((m, n) ) (4, 1); (4, 2); (4, 5); (16, 15); (28, 15); (44, 15); (28, 27)) clusters were investigated using generalized gradient approximation (GGA) and the Hubbard U (GGA+U) method within density functional theory (DFT). It is found that the electronic structures of bulk Cu2O calculated by the GGA and GGA+U are similar. The structures of CumOn ((m, n) ) (4, 1); (4, 2); (4, 5)) are all planar. For the bulk-product CumOn ((m, n) ) (16, 15); (28, 15); (44, 15); (28, 27)), O atoms prefer to be the outermost atoms. We classified two types of clusters on the basis of their O to Cu atomic ratios. One is O-rich clusters, i.e., Cu4O5, Cu16O15, and Cu28O27. The other is O-poor clusters, i.e., Cu4O, Cu4O2, Cu28O15, and Cu44O15. The calculation results show that the O-rich clusters have longer average Cu-Cu bonds and larger binding energy than those of the O-poor ones. More interestingly, the former are magnetic and give ferromagnetic ordering while the latter are nonmagnetic. The hydrogenation of O-terminated clusters can improve its stability but suppress its magnetism. The study may be extremely useful for the potential applications of Cu2O nanopaticles in the catalysis and semiconductor fields. 1. Introduction During the past decades, metal oxides, especially transitionmetal oxide nanoparticles, have attracted more and more attention because of their wide applications in catalysis, fuel cell, sensor technology, magnetic materials, etc. The structures of small clusters containing a few to hundreds of atoms are different from bulk materials. In an atomic cluster, most of the atoms are located at the surface. Its structural crossover phenomena and electronic confinement effects1 determine physical and chemical properties of materials. With the increase of cluster sizes, more atoms can be considered to share the same characteristics and be positioned as that of solid material. In the center of large clusters, condensed phases are formed in correspondence to bulk modification.2 Cu2O is of great interest in catalysis3 and semiconductor fields due to its material abundance, preparation simplicity, nontoxicity, and suitable band gap. Its p-type conduction property and broad response to visible light (VL) suggest that one of the most important applications of Cu2O is its function as a component in a photovoltaic solar cell. It was found that by combining Cu2O with an n-type semiconductor, such as TiO24 or ZnO,5 * To whom all correspondence should be addressed. E-mail: yuying@ phy.ccnu.edu.cn. Fax: 86-27-67861185. † Huazhong Normal University. ‡ Zhengzhou University. § Nanjing University. | University of Science and Technology Beijing. ⊥ Wuhan Institute of Technology. # Virginia Commonwealth University.
the new composite materials show good energy conversion efficiency, implying that it is an economical candidate to replace the traditional high-cost crystalline silicon solar cell. In addition, it is also a promising material able to be applied in developing transparent conducting electronic devices by alloying with alkaline-earth oxide.6 As a catalyst, Cu2O has been found to degrade organic pollutants in wastewater under visible light.7-12 When catalysts hybridize with magnetic materials such as Fe, it is believed to benefit the recycling of the catalysts in solution.13 If Cu2O has magnetic properties as well, it will become an even better photocatalyst. With the growing interest in the new and exciting field of “oxide spintronics”, magnetic oxide semiconductors are attracting more attention as a potential candidate for materials in spintronics-related technology.14,15 More recently, there is new evidence indicating that Cu2O nanoparticles (∼10-100 nm) surprisingly have a native ferromagnetic property.16,17 In our group, we have succeeded in synthesizing a series of Cu2O nanomaterials including nanowhiskers,7 nanorods,8 nanoarchitectures,9 aligned nanosheets arrays,10 nanoflowers,11 and nanoparticles.12 They are composed of super small Cu2O particles within the size range of ∼2-8 nm. However, the magnetic properties of the tiny Cu2O nanoparticles (∼2-8 nm) demonstrate the paramagnetic characteristics shown in Figure S1 (Supporting Information). Oxygen concentration and particle size are proposed to be related to the magnetic properties.16 Bulk native point defects and surface properties from theoretical studies are indicated to be the origin of the Cu2O native magnetism.14 However, the kind of study has not been consid-
10.1021/jp103703n 2010 American Chemical Society Published on Web 07/23/2010
8418
J. Phys. Chem. A, Vol. 114, No. 32, 2010
Yang et al.
ered due to the complexity of the electronic structure, which would require post-DFT approaches, such as DFT+U.14 Therefore, it remains to investigate the puzzling magnetic properties relative to the sizes and compositions of those cluster materials further. Moreover, the study of CumOn clusters could also shed light on our present knowledge of Cu2O photocatalysis and magnetic properties as well as highlight the application of cuprous oxide-based nanomaterials. The small spherical nanoparticles of Cu2O synthesized in our group have a cuprite crystal structure.7-12 Therefore, the clusters under investigation here are truncated from bulk Cu2O. Bulk Cu2O has been well studied using first principles methods including Hartree-Fock,18,19 DFT,6,19-24 etc. Efforts have also been devoted to understanding the geometries and electronic structures of CuxOy clusters.25-29 However, partly due to the complexity in structure, previous theoretical efforts mainly focused on building blocks in small clusters CuxOy (x ) 1-2, y ) 1-6). Studies on larger CumOn clusters, which are more relevant to applications, are still rare. In this work, we present our computational results on the structural and electronic properties of a series of CumOn clusters ranging from 5 to 59 atoms. Because the (-OH) group was detected in the samples of Cu2O nanoparticles in our experiment,12 the hydrogenation of the O-terminated cluster is also calculated to check the hydrogenation effect on the cluster. We have considered both O- and Cu-terminated clusters with different oxygen concentrations. Furthermore, both bulk Cu2O and clusters are calculated using density functional theory (DFT)30 with GGA-PW91 approaches. Considering the nonstoichiometricity of clusters, especially the O-rich oxides that induce the Cu d orbitals to be partially occupied, the DFT+U approach is also used for all the bulk and clusters calculations. The results obtained by the two methods are compared. 2. Computational Details The calculations were performed using the Vienna ab initio simulation package (VASP)31,32 based on self-consistent density functional theory (DFT).30 We used a frozen-core projector augmented wave (PAW) approach33,34 to describe core-valence interaction and Perdew and Wang (PW91)35-type GGA for the exchange-correlation functional. To explore strong on-site Coulomb repulsion among localized Cu 3d electrons, GGA+U formalism given by Dudarev et al.36 was used,
ELSDA+U ) ELSDA +
(U - J) 2
[( ) )] (∑
∑ ∑ nmσ ,m σ
m1
m1,m2
1
-
1
nˆmσ 1,m2nˆmσ 2,m1
(1)
where the expression in square brackets is the total number of d electrons Nσ, while U and J are respectively the spherically averaged matrix elements of screened Coulomb and exchange energy. σ represents spin and m is the magnetic quantum number of the d orbital. Since only the difference (U - J) is relevant, we have treated them as one single parameter labeled as U. To achieve appropriate parameters for cluster calculation, our calculations began with bulk Cu2O. We used a plane wave cutoff energy of 508 eV for all calculations to ensure accurate results. For the cubic bulk unit cell Cu2O, we used a 7 × 7 × 7 Monkhorst-Pack k-point sampling scheme38 with 20 irreducible k-points. For the density of states (DOS) calculation, k-points were performed using tetrahedron method37 with a 15 × 15 × 15 Monkhorst-Pack gird37 (120 irreducible k-points). A smear-
TABLE 1: Structural Parameters of Bulk Cu2O Calculated by GGA-PW91 (G-P), GGA+U (G+U), Hatree-Fock, and GGA-PBE Methodsa lattice (Å) Cu-O (Å) Cu-Cu (Å)
G-P
G+U
HFb
PBEc
Expd
4.30 1.86 3.04
4.26 1.85 3.01
4.28 1.85 3.03
4.34/4.29 1.88/1.86 3.07
4.27 1.85 3.01
a The experimental values are also listed for comparison. b From ref 19. c From refs 6 and 22. d From ref 40.
ing width of σ was tested to 0.01, which was enough for all the calculations. For the GGA+U method, which can partially improve the prediction of band gap, U ) 7 was applied on Cu 3d electrons. The same U was also used for the calculations of all the clusters. The geometry of the unit cell (ionic coordinates and lattice parameters) was optimized without symmetry constraint. Cluster calculations were performed in a supercell with sufficient vacuum regions to minimize the interaction between neighboring clusters caused by periodic boundary conditions. The ground states of small clusters were obtained by comparing the energies of many proposed configurations optimized by simulated annealing and conjugated gradient (CG)39 methods. For a large number of isomers, the initial structure of large clusters was generated from bulk Cu2O. The average binding energy per atom in the cluster is defined as
Eb )
-(ECumOnHp - mECu - nEO - pEH) (m + n + p)
(2)
where ECumOnHp denotes the total energy of CumOnHp clusters, and ECu, EO, and EH represents the energy of free Cu, O, and H atom, respectively. 3. Results and Discussion 3.1. Bulk Cu2O. Cu2O has a cuprite structure with symmetry group Pn3jm, #224. The lattice constant given by the GGAPW91 functional is 4.30 Å, about 0.8% larger than experimental value, 4.27 Å.40 The calculated Cu-O bond length is 1.86 Å and the nearest Cu-Cu distance is 3.04 Å. By comparison, the experimental values are 1.85 and 3.01 Å, respectively. As can be seen, the calculated data match the experimental data very well. To evaluate the performance of GGA in this system, we compare the results obtained by GGA-PW91, GGA+U, and other theoretical calculations with experimental data in Table 1. It can be seen that GGA-PW91 and GGA+U provide a good description for bulk Cu2O. The calculated total density of states (TDOS) and the partial density of states (PDOS) are presented in Figure 1, where the Fermi-level is set to zero. Solid and dashed lines correspond to the results calculated by GGA-PW91 and GGA+U, respectively. It can be seen that the Cu 3d levels extend from -5 to 0 eV, hybridized with Cu 4s and O 2p states at the top of the valence band (VB). From -7.5 to -5 eV, O 2p mixed with Cu 3d states predominantly occupy the region. The bottom of the conduction band (CB) is derived from Cu 4s and Cu 3d. Furthermore, the peaks from the GGA+U calculation slightly shift the VB downward and the CB upward. A comparison of our calculated photoemission peaks with experimental values41 is shown in Table 2. Generally, our computational values are in good agreement with the experimental values.
Magnetic Properties of CumOn Clusters
J. Phys. Chem. A, Vol. 114, No. 32, 2010 8419
Figure 1. DOS and PDOS for bulk Cu2O calculated by GGA-PW91 (solid line) and GGA+U (dashed line). For PDOS, notice that the different panels have different scales on the y axis.
TABLE 2: Some Energy Levels Calculated by GGA-PW91 and GGA+U for Bulk Cu2O (in eV) in Comparison with Experimental Observationsa DFT-PW91 GGA+U experimentb
A
B
C
D
E
F
-0.59 -0.61 -1.01
-1.21 -1.56 -2.00
-1.72 -2.81 -2.95
-2.21 -3.31 -3.37
-5.45 -5.03 -6.04
-6.97 -6.65 -7.25
a A, B, C, D, E, and F are the corresponding energy peaks from Figure 1, respectively. b From ref 41.
Figure 2. Band structures for Cu2O: (a) calculated by the GGA+U approach with a band gap of 0.80 eV and (b) calculated by the GGAPW91 method with a band gap of 0.51 eV. The Fermi level is set to zero.
The calculated band structures of bulk Cu2O using GGA+U (panel a) and GGA-PW91 (panel b) are shown in Figure 2. The direct band gap at Γ-point is 0.51 eV using DFT-PW91. The calculated band gap agrees well with 0.52 eV using LDA42 and 0.47 eV using DFT-PBE.22 However, it is significantly lower than the experimental value 2.0 eV due to a well-known DFT error. The band gap increases to 0.80 eV using GGA+U, as shown in Figure 2a. Because of the closed 3d shell, Cu2O is not considered to be a strongly correlated material.24 The GGA+U method partially improves the accuracy of band gap calculation, although the improvement is not pronounced. On the basis of the above bulk results, we began our theoretical cluster work. For nonstoichiometric clusters, especially the O-rich copper clusters that induce the Cu d states to be partially occupied, the DFT+U approach is also necessary and used for cluster calculations. 3.2. Clusters of CumOn. We generated six O-centered spherical clusters of Cu4O, Cu4O5, Cu16O15, Cu28O15, Cu44O15, and Cu28O27 from cubic bulk Cu2O. Both Cu terminations and O concentrations are taken into account. Figure 3a shows a schematic picture of how to cut the O-centered Cu2O spherical clusters from the bulk. Cu16O15 (radii R ) 4.30 Å), Cu28O15 (R ) 4.69 Å), and Cu44O15 clusters are terminated by Cu atoms and possess low O weight percentages. Cu4O5, Cu16O15, and Cu28O27 are truncated by O atoms and have high O weight
Figure 3. (a) Schematic picture of how to cut the O-centered Cu2O spherical clusters from bulk. (b) Optimization geometry for the atomic small clusters studied: I, Cu4O; II, Cu4O2; III, Cu4O5. Cu atoms: green spheres. O atoms: red spheres.
percentages. Because it is difficult to obtain clusters with high O weight percentages, Cu28O27 is made by removing some outer Cu atoms. 3.2.1. Cu4On (n ) 1, 2, 5). We considered a large number of isomers as reasonable initial structures for Cu4O and Cu4O5 clusters. Additionally, to investigate the relationship between oxygen concentration and cluster properties, stoichiometric cluster Cu4O2 is also considered. All the isomers of the three clusters we have studied tend to form 2D structures. The stable structures of Cu4O, Cu4O2, and Cu4O5 are shown in Figure 3b. All of them have C1 point symmetry. The bond length and bond angles of these clusters are similar to those of reported copper oxide cluster calculations.25-29,43 For (I) Cu4O, 4 copper atoms form a dense configuration and tightly bond to each other, all the Cu-Cu bond lengths are shorter than that of bulk metal copper, 2.540 Å in experiment. The average Cu-Cu bond length is 2.395 Å and the O atom joins two copper atoms for the formation of a (Cu2O) unit. For (II) Cu4O2, the tightly bonded 4 copper atoms are disturbed by inserting two O atoms to form two (Cu2O) units. This structure is in accordance with ref 43. For (III) Cu4O5, the tightly bonded 4 copper framework is thoroughly destroyed. One O atom inserts into the center of copper framework and the other 4 O atoms attach to two copper atoms for the formation of (Cu2O) units (Figure 3b). From this result, it can be concluded that the (Cu2O) unit plays a very important role in small cluster formation, which agrees well with other cluster studies.27-29,43 Semiconductor clusters can also possess considerably different electronic properties from the bulk owing to the quantum confinement effect. The energy gap between the highestoccupied molecular orbital (HOMO) and the lowest-unoccupied molecular orbital (LUMO) plays a very important role to characterize cluster properties. To study the electronic properties of the clusters in detail, we performed spin polarized calculations and plotted their electronic density of states in Figure 4 (Ι, Cu4O; II, Cu4O2; III, Cu4O5). The band gaps for the three clusters using GGA-PW91 are 0.70, 0.50, and 0.50 eV, respectively. The band gap becomes wider when the O concentration decreases. Besides, no magnetism is found for these small clusters by GGA-PW91. Using the GGA+U method, the band gap is increased to 1.12 and 1.09 eV for Cu4O and Cu4O2, respectively, and again no magnetism is found. However, for Cu4O5, the spin-
8420
J. Phys. Chem. A, Vol. 114, No. 32, 2010
Figure 4. DOS plot for the clusters studied: I, Cu4O; II, Cu4O2; III, Cu4O5. The Fermi level is set to zero; the solid and dashed lines are the results from GGA-PW91 and GGA+U.
up and spin-down plot is split with a gap of 1.28 eV for spinup and 0.96 eV for spin-down. The local spin moment on a Cu atom can go up to 0.29 µB and on an O atom be slightly polarized to 0.15 µB, leading to ferromagnetism in Cu4O5 with a total spin moment of 2 µB. 3.2.2. Large Clusters CumOn. Considering a large number of isomers for the larger clusters, we focused on bulk-generated clusters for the remaining clusters. After relaxation, (a) Cu16O15 and (b) Cu28O15 clusters have Td space group, (c) Cu44O15 has C3V point symmetry, and (d) Cu28O27 has C1 point symmetry. The initial and optimized structures of the large clusters are shown in Figure 5. Geometric optimization shows that the O-rich clusters are different from the O-poor clusters. For the O-rich clusters, the outermost atoms remain O-terminated. For (a) Cu16O15, the surface atoms form new Cu-O and Cu-Cu bonds, which makes the cluster exhibit moderate contraction. The average Cu-O bond length dCu-O is 1.79 Å (GGA+U: 1.77 Å), showing -3.76% (GGA+U: -4.84%) variation compared to the bulk. The average Cu-Cu bond length dCu-Cu is 2.85 Å (GGA+U: 2.81 Å), which is longer than the Cu-Cu distance 2.55 Å in Cu metal. For (d) Cu28O27, the cluster undergoes a drastic contraction because of the copper vacancies resulting from removing Cu atoms at the outer surface. The formation of new Cu-O and Cu-Cu bonds is observed at the outermost surface and drives the cluster’s drastic reconstruction. The average dCu-Cu is 2.68 Å (GGA+U: 2.70 Å). For the O-poor clusters, (b) Cu28O15 and (c) Cu44O15, drastic structure reconstruction is exhibited at the outermost surface of the cluster, in which Cu atoms move inward while O atoms become the outermost atoms of the clusters. After relaxation, the new Cu-Cu and Cu-O bonds at the cluster surface are formed and, finally, O atoms become the outmost surface atoms for all the large clusters. The large displacement of the atoms at the surface is due to the large electronic affinity of O atoms and the tendency to form Cu-O covalent bonds, which drives the Cu atoms at the surface to move deeper into cluster. Hence,
Yang et al.
Figure 5. Structure of large clusters studied. The middle and right columns show each nanocluster in its initial and relaxation configurations, respectively. (a) Cu16O15 and (d) Cu28O27 possess high O/Cu ratios. (b) Cu28O15 and (c) Cu44O15 possess low O/Cu ratios. Cu atoms: green spheres. O atoms: red spheres.
Figure 6. DOS plot for the clusters studied. (a) Cu16O15 and (d) Cu28O27 possess high O/Cu ratios. (b) Cu28O15 and (c) Cu44O15 possess low O/Cu ratios. The Fermi level is set to zero; the solid line stands for the GGAPW91 result.
an O-rich surface is formed. The different behavior of the oxygen group element and metal atoms is similar to that of other semiconductor clusters,44-46 especially the CdS nanocrystal.47 The average dCu-Cu for the two clusters is 2.58 and 2.57 Å (GGA+U: 2.54 Å), respectively. For bulk-product CumOn ((m, n) ) (16, 15); (28, 15); (44, 15); (28, 27)), O atoms prefer to be the outermost too. The two types of clusters present different electronic properties. The O-rich clusters (a) Cu16O15 and (d) Cu28O27 are found to be magnetic, while the O-poor clusters are nonmagnetic. We performed spin polarized calculations, and the resulting electronic density of states is shown in Figure 6. For (a) Cu16O15 and (d) Cu28O27, the majority and minority spin DOS splits. The PDOS of the cluster (a) (not given here) shows that the
Magnetic Properties of CumOn Clusters
J. Phys. Chem. A, Vol. 114, No. 32, 2010 8421
TABLE 3: Properties of CumOn Clusters Calculated by GGA-PW91 (G-P) and GGA+U (G+U)a dCu-Cu (Å) bulk Cu2O Cu4O Cu4O2 Cu4O5 Cu16O15 Cu28O15 Cu44O15 Cu28O27
dCu-O (Å)
Eb (eV)
M (µB)
G-P
G+U
G-P
G+U
G-P
G-P
G+U
3.04 2.39 2.38 2.75 2.85 2.58 2.57 2.68
3.01 2.40 2.32 2.71 2.81 2.54 2.53 2.70
1.86 1.78 1.86 1.84 1.79 1.84 1.97 1.95
1.85 1.76 1.84 1.83 1.77 1.85 1.97 1.94
2.274 2.562 3.413 3.461 3.477 3.354 3.713
0 0 0 0 14 0 0 3.95
0 0 0 2 14 0 0 22
a Eb (eV) represents the average binding energy of CumOn clusters. dCu-Cu (Å) and dCu-O (Å) are the average Cu-Cu and Cu-O bond length and the M (µB) denotes the magnetization.
calculated results in detail are shown in Table 3. In this table, dCu-Cu (Å) and dCu-O (Å) are the average bond length of Cu-Cu and Cu-O. Eb (eV) denotes the average binding energy of the clusters and M is the magnetic moment. The paramagnetic properties of the tiny Cu2O nanoparticles (∼2-8 nm) are shown in Figure S1 (Supporting Information). Since the (-OH) group exists in Cu2O nanoparticles,17 a hydrogenated O-terminated cluster Cu16O15H4 is also investigated to check the effect of hydrogenation on the clusters. It is found that the binding energy increases to 3.608 eV, which is larger than that for the non-hydrogenated ones. The total magnetism is reduced to 10 µB. The effective reduction of the clusters by hydrogenation results in weakened magnetism. Overall, hydrogenated clusters can improve stability and weaken magnetism. Therefore, hydrogenation and oxygen concentration in the clusters are responsible for adjusting their magnetic properties. This would be why the Cu2O nanoparticles we experimentally prepared exhibit paramagnetic properties while theoretical ones reported ferromagnetic properties.10,11 4. Conclusions
Figure 7. (a) Average Cu-Cu bond length of the clusters calculated by GGA-PW91 and GGA+U method and (b) average binding energy of clusters.
states around Fermi level for minority spin result from hybridization between Cu 3d and O 2p states. It is interesting to find that the PDOS obtained from GGA+U almost have no significant change, indicating the delocalized properties of the states. These states cut through the Fermi level and thus contribute to electrical conductivity. Both O and Cu atoms are all polarized in the two clusters. For Cu16O15, the largest local spin moment goes up to 0.75 µB on O and 0.50 µB on Cu, leading to a ferromagnetic ordering with a total spin moment of 14 µB. As for Cu28O27, the spin polarization is localized on its surface atoms. The largest local spin moment on the Cu atom is 0.23 µB and on the O atom is 0.20 µB, resulting in a ferromagnetic ordering with total spin moment of 3.95 µB. On the other hand, for (b) Cu28O15 and (c) Cu44O15 clusters, there are a few empty states, mainly derived from the 2p states at the surface, which is just above the Fermi level (Figure 6). As the spin-up and spin-down states are almost equal, the calculated magnetic moment is zero. The calculation results from GGA+U are also similar to those from GGA-PW91. There is only the improvement of the magnetic values as listed in Table 3. The reason is that the magnetic properties are induced by the hybridization between Cu 3d and O 2p at the surfaces of the clusters and the exchange-correlated effects of d electrons become very strong for the O-rich clusters. Comparing the average dCu-Cu of all the studied clusters (Figure 7a), it can be found that the O-rich clusters (Cu4O5, Cu16O15, and Cu28O27) have longer average Cu-Cu bonds than the O-poor clusters, implying that O atoms act as the spacers in the O-rich clusters. GGA+U gives similar results for all the clusters. The average binding energy Eb is shown in Figure 7b. It can be seen that the O-rich clusters (Cu4O5, Cu16O15, and Cu28O27) have larger average binding energies Eb. This indicates that the formation of copper vacancy is obviously much easier than that of stoichiometric (bulk) and O vacancy, implying an even more effective catalytic ability of these clusters. The
First principles can be used to investigate the structural and electronic properties of a series of CumOn clusters. GGA and GGA+U methods lead to similar results for Cu2O. The O-rich clusters (Cu4O5, Cu16O15, and Cu28O27) have longer average Cu-Cu bond lengths and higher stabilities. Due to the size and geometrical effects, the formation of copper vacancy (oxidation process) is obviously much easier, implying even more effective catalysis of these clusters. Moreover, these O-rich clusters are magnetic. When the clusters are hydrogenated, the magnetism can be reduced and the stability of the clusters can be improved further. This study will give a new understanding of the CumOn clusters, facilitate further investigation of CumOn materials, and guide the controllable preparation of cuprous oxide nanomaterials. Acknowledgment. This work is financially supported by the National Natural Science Foundation of China (Nos. 20973070 and 50804035), the Key Project of Chinese Ministry of Education (No. 109116), National Basic Research Program of China (No. 2009CB939704), the Natural Science Foundation of Hubei Province, China (No. 2009CDB004), and the Program of Introducing Talents of Discipline to Universities (No. B08033). Supporting Information Available: Figure S1 showing the ZFC/FC magnetization as a function of temperature. This material is available free of charge via the Internet at http:// pubs.acs.org. References and Notes (1) Gruner, M. E.; Rollmann, G.; Entel, P.; Farle, M. Phys. ReV. Lett. 2008, 100, 087203-087209. (2) Degimann, P.; Ahlrichs, R.; Tsereteli, K. J. Chem, Phys. 2002, 116, 1585–1587. (3) Poizot, P.; Laruelle, S.; Grugeon, S.; Dupront, L.; Taracon, J. M. Nature 2000, 407, 496–499. (4) Siripala, W.; Ivanovskaya, A.; Jaramillo, T. F.; Baeck, S. H.; McFarland, E. W. Sol. Energy Mater. Sol. Cells 2003, 77, 229–237. (5) Jeong, S. S.; Mittiga, A.; Salza, E.; Masci, A.; Passerini, S. Electrochim. Acta 2008, 53, 2226–2231. (6) Nolan, M.; Elliott, S. D. Phys. Chem. Chem. Phys. 2006, 8, 5350. Nolan, M.; Elliott, S. D. Thin Solid Film. 2008, 516, 1468–1472. (7) Yu, Y.; Du, F. P.; Yu, J. C.; Zhuang, Y. Y.; Wong, P. K. J. Solid State Chem. 2004, 177, 4640–4647. Yu, Y.; Huang, W. Y.; Du, F. P.; Ma, L. L. Mater. Sci. Forum 2005, 3531, 475–479. (8) Ma, L. L.; Peng, M.; Li, J. L.; Yu, Y.; Chen, Z. H. Proc, IEEE Int. Conf. Nanotechnol. (Hong Kong), 7th 2007, 975–978.
8422
J. Phys. Chem. A, Vol. 114, No. 32, 2010
(9) Luo, Y. S.; Li, S. Q.; Ren, Q. F.; Liu, J. P.; Xing, L. L.; Wang, Y.; Yu, Y.; Jia, Z. J.; Li, J. L. Cryst. Growth Design 2007, 7 (1), 87–92. (10) Ma, L. L.; Lin, Y. L.; Wang, Y.; Li, J. L.; Wang, E. K.; Qiu, M. Q.; Yu, Y. J. Phys. Chem. C 2008, 112, 18916–18922. (11) Ma, L. L.; Li, J. L.; Sun, H. Z.; Wang, E. K.; Yu, Y. Mater. Res. Bull. 2010, 45, 961–968. (12) Yu, Y.; Ma, L. L.; Huang, W. Y.; Li, J. L.; Wong, P. K.; Yu, J. C. J. Solid State Chem. 2005, 178, 1488–1492. (13) Ai, Z. H.; Lu, L. R.; Li, J. P; Zhang, L. Z.; Qiu, J. R; Wu, M. H. J. Phys. Chem. C, 2007, 111, 7430–7436. (14) Soon, A.; Cui, X. Y.; Delley, B.; Wei, S. H.; Stampfl, C. Phys. ReV. B 2009, 79, 035205-035211. (15) Sieberer, M.; Redinger, J.; Mohn, P. Phys. ReV. B 2007, 75, 035203-035209. (16) Yermakov, A. Y.; Uimin, M. A.; Mysik, A. A.; Vykhodets, V. B.; Kurennykh, T. E.; Sokolov, V. I.; Gaviko, V. S.; Schegoleva, N. N.; Gruzdev, N. B. J. Magn. Magn. Mater. 2007, 310, 2102–2104. (17) Sokolov, V. I.; Yermakov, A. Y.; Uimin, M. A.; Mysik, A. A.; Vykhodets, V. B.; Kurennykh, T. E.; Gaviko, V. S.; Schegoleva, N. N.; Gruzdev, N. B. J. Exp. Theor. Phys. 2007, 105, 65–67. (18) Laskowski, R.; Blaha, P.; Schwarz, K. Phys. ReV. B 2003, 67, 075102-075109. (19) Ruiz, E.; Alvarez, S.; Alemany, P.; Evarestov, R. A. Phys. ReV. B 1997, 56, 7189–7196. (20) Ghisjen, J.; Tjeng, L. H.; Elp, J.; van; Eskes, H.; Westerink, J.; Sawatzky, G. A.; Czyzyk, M. T. Phys. ReV. B 1988, 38, 11322–11330. (21) Raebiger, H.; Lany, S.; Zunger, A. Phys. ReV. B 2007, 76, 45209– 45215. (22) Soon, A.; Sa¨ohnel, T.; Idriss, H. Surf. Sci. 2005, 579, 131–140. (23) Boudin, S.; Felser, C.; Studer, F. Solid State Sci. 2003, 5, 741– 744. (24) Bruneval, F.; Vast, N.; Reining, L. Phys. ReV. Lett. 2006, 97, 267601–267608. (25) Pouillion, Y.; Massobrio, C. Chem. Phys. Lett. 2002, 356, 469– 475.
Yang et al. (26) Baruah, T.; Zope, R. R.; Pederson, M. R. Phys. ReV. A 2004, 69, 023201-023108. (27) Dai, Bing; Tian, Li; Yanga, Jinlong. J. Chem. Phys. 2004, 120, 2746–2751. (28) Massobrioa, C.; Pouillon, Y. J. Chem. Phys. 2003, 119, 8305–8310. (29) Wang, L. S.; Wu, H. B.; Desai, S. R. Phy. ReV. B 1996, 53, 8028– 8031. (30) Kohn, W.; Becke, A. D.; Parr, R. G. J. Phys. Chem. 1996, 100, 12974–12980. (31) Kresse, G.; Hafner, J. Phys. ReV. B 1993, 48, 13115–13118. (32) Kresse, G.; Furthmu¨ller, J. Phys. ReV. B 1996, 54, 11169–11186. (33) Blo¨chl, P. E. Phys. ReV. B 1994, 50, 17953–17979. (34) Kresse, G.; Joubert, J. Phys. ReV. B 1999, 59, 1758–1775. (35) Perdew, J. P. In Electronic Structure of Solids; Ziesche, P., Eschrig, H., Eds.; Verlag, Akademie: Berlin, 1991. (36) Dudarev, S. L.; Botton, G. A.; Savrasov, S. Y.; Humphreys, C. J.; Sutton, A. P. Phys. ReV. B 1998, 57, 1505–1509. (37) Monkhorst, H. J.; Pack, J. D. Phys. ReV. B 1976, 13, 5188–5192. (38) Blo¨chl, P. E.; Jepsen, O.; Andersen, O. K. Phys. ReV. B 1994, 49, 16223–16233. (39) Teter, M. P.; Payne, M. C.; Allan, D. C. Phys. ReV. B 1989, 40, 12255–12263. (40) Werner, A.; Hocheimer, H. D. Phys. ReV. B 1982, 25, 5929–5934. (41) Shen, Z. X.; et al. Phys. ReV. B 1990, 42, 8081–8085. (42) Nie, X.; Wei, S. H.; Zhang, S. B. Phys. ReV. B 2002, 65, 075111075117. (43) Jadraque, M.; Martı´n, M. Chem. Phys. Lett. 2008, 456, 51–54. (44) Pal, S.; Goswami, B.; Sarkar, P. J. Chem. Phys. 2005, 123, 044311044317. (45) Sarkar, P.; Springborg, M. Phys. ReV. B 2003, 68, 235409–235415. (46) Roy, S.; Springborg, M. J. Phys. Chem. B 2003, 107, 2771–2779. (47) Junkermeier, C. E.; Lewis, J. P.; Bryant, G. W. Phys. ReV. B 2008, 77, 205125–205131.
JP103703N
