J Mol Model (2014) 20:2094 DOI 10.1007/s00894-014-2094-y
ORIGINAL PAPER
First-principles calculations of nickel, cadmium, and lead adsorption on a single-walled (10,0) carbon nanotube Mirele Bastos & Ihosvany Camps
Received: 11 August 2013 / Accepted: 25 November 2013 / Published online: 11 February 2014 # Springer-Verlag Berlin Heidelberg 2014
Abstract The adsorption of Ni, Cd, and Pb on a zigzag (10, 0) carbon nanotube (CNT) surface was investigated using density functional theory. Binding energy calculations were performed, and the results indicated that the three metals are stably adsorbed on the nanotube surface. Moreover, the results showed that Cd is physisorbed whereas Ni and Pb are chemisorbed. Our studies show that the electronic properties of the CNT are modified by the chemisorption mechanism (Ni and Pb). After Ni and Pb adsorption, the nanotube changes from being a semiconductor to a metallic conductor. The nanotube remains semiconductive upon Cd physisorption, although a decrease in the band gap is observed. Also, Ni or Pb adsorption triggers a change in the magnetism of the nanotube through the induction of spin polarization. Not only can these results of our calculations be used to explain the adsorption mechanisms of these heavy metals on the CNT, but they are also useful for evaluating the potential of carbon nanotubes (CNTs) to act as filters and sensors of such metals. Keywords Nanostructures . Ab initio calculations . Electronic structure . Band structure
Introduction In recent decades, technological and industrial advances and a lack of care in toxic material disposal have resulted in the gradual extermination of many plant and animal species [1]. The pollutants that cause the most concern include heavy metals, such as Ni, Cd, Pb, Cr, Cu, Hg, Zn, and Mn. There M. Bastos : I. Camps (*) Laboratório de Modelagem Computacional—LaModel, Instituto de Ciências Exatas, Universidade Federal de Alfenas—Unifal-MG, CEP 37130-000 Alfenas, Minas Gerais, Brazil e-mail: [email protected]
is great apprehension over the exposure of populations to heavy metals because they are not biodegradable and they have long biological lifetimes, which hinder their removal from the body [2]. One way to circumvent this problem is to use solid-phase materials that retain heavy metals and are simple and inexpensive. Such a material should have a large surface area, a small size, and high chemical stability. Among the materials that exhibit such characteristics are carbon nanotubes (CNTs) [3–10]. A search for new materials with high adsorption capacities and surface areas suggests that CNTs are strong candidates. Many theoretical and experimental studies have attempted to unravel the interaction mechanisms of metals [11, 12], gases [6, 13–17], and organic molecules [18, 19] with pure CNTs (with various geometries), CNTs doped with additional elements [14–17, 20], and CNTs decorated with organic functional groups [12, 18, 21, 22]. Experimental studies on singlewalled carbon nanotube field-effect transistors have already been reviewed [6, 19]. However, CNTs that interact with metals, especially transition metals, have been studied for other applications too, such as in nanodevices [15, 23], spintronics [24–27], and metal-catalytic systems [28, 29]. In the work described in the present paper, we studied the interactions between the heavy metals Ni, Cd, and Pb and a single-walled CNT. We chose these metals due to the growth of industrial activities such as the burning of coal and oil (for power generation), as these activities are contributing to the release of Ni, Cd, and Pb into the environment. Furthermore, there is no comprehensive study in the literature on the interaction of these metals with a single-walled CNT. We used density functional theory to calculate the differences in the electronic properties of the CNT before and after metal adsorption. This paper is organized as follows. In the next section (“Modeling details”), we describe the methodology and parameters used for the calculations and the system configuration employed. Subsequent sections discuss results relating to
2094, Page 2 of 8
J Mol Model (2014) 20:2094
adsorption (“Ni, Cd, and Pb adsorption at the CNT surface”) and the electronic properties of the nanotube (“Electronic properties”). In the final section of the paper (“Conclusions”), we draw conclusions from the results we obtained.
Modeling details We used density functional theory (DFT) as implemented in the SIESTA code [30] and the local density approximation (LDA) for the exchange-correlation potential. Both the Ceperley–Alder (CA) [31] scheme and the Perdew and Zunger parameters [32] (PZ) were used. To avoid the need to explicitly treat the core electrons, the norm-conserving Troullier–Martins pseudopotentials [33] were used. For the valence electrons, we used a split-valence double-zeta basis set with polarization functions (DZP) [34]. Periodic boundary conditions were applied using the supercell approach. A (10, 0) zigzag pure CNT with 120 atoms, or the same nanotube plus a heavy metal (121 atoms), was placed (oriented in the z-direction) inside a 38×25×12 Å cell. Using this configuration, each system image was separated by at least ∼10 and ∼22 Å in the x- and y-directions, respectively, to avoid interactions between two neighboring systems. To produce sufficiently accurate results, convergence studies were performed for the grid (mesh) cutoff energy and the number of k-points. Total energy convergence was possible for the systems tested here using a mesh cutoff of ∼200 Ry; however, we used 300 Ry to ensure accuracy. To estimate the required number of k-points, the Brillouin zone was sampled in accordance with Monkhorst and Pack [35]; convergence was obtained for a 1×1×8 set, but a 1×1×11 k-point sampling set was actually used in this work. The electronic properties were calculated for systems with fully relaxed atomic coordinates obtained by performing geometry optimization until the Hellmann–Feynman forces were below 0.03 eV Å −1. For the adsorption studies, the CNT+metal system energy was calculated starting with a manually specified metal position (supplied by an input file). The binding energy (Eb) for each heavy metal adsorbed on the CNT was calculated using the following expression [36]: E b ¼ E CNTþM −ECNT −EM −E CC ;
ð1Þ
where ECNT and EM are the Kohn–Sham energies for the CNT and metal, respectively, ECNT+M is the Kohn–Sham energy of the complex CNT+metal, and ECC is the counterpoise correction that accounts for the effects of the basis set superposition error (BSSE), as follows: E CC ¼ E ghost−M þ Eghost−CNT ;
ð2Þ
where Eghost−M and Eghost−CNT are the metal and CNT Kohn– Sham energies, respectively, and the atoms were replaced by ghost atoms [30]. The atomic charge distribution can be calculated from first principles. In this work, we examined the topological charge density distribution using the Mulliken population analysis [37, 38] implemented in the SIESTA code. Variations in the charge on the metal due to transfer between the CNT and the metal were calculated using the following expression: iso ΔQM ¼ Qabs M −QM ;
ð3Þ
where Qabs M is the total charge on the metal after adsorption and Qiso M is the total charge on the isolated metal.
Results Ni, Cd, and Pb adsorption at the CNT surface In the work reported here, we were interested in the interaction between a heavy metal (either Ni, Cd, or Pb) and a (10, 0) zigzag CNT. The CNT surface has different areas where the metal can interact. The physical origin of these different adsorption sites is mainly the shape of the overlap between the frontier atomic orbitals of carbon atoms (from the CNT) at each site and the frontier orbitals of the adsorbed atoms. Other factors that can affect the adsorption include the electronic distribution and charge polarization effects on the CNT sites and on the added metal. It is possible to calculate and plot the frontier orbitals in those regions. Following the results of Durgun et al. [39, 40], we only studied the interaction of each metal with the most probable binding site: the bridge site. In these works, the authors mainly calculated the binding energies of several metals adsorbed on a CNT. In terms of the electronic properties, they only showed the electronic bands after the adsorption of an Al, C, Si, or Ti atom (none of which are studied in the present work). Thus, in contrast to the works of Durgun et al. [39, 40], the present work represents a more detailed study of the adsorption of Ni, Cd, and Pb on a CNT, including a deep exploration of the mechanisms involved and the resulting modifications to the electronic structure of the CNT. To accomplish this, the binding energies were calculated using a more complete theory: the counterpoise correction, which accounts for the effects of the basis set superposition error (BSSE). This correction is used to avoid under/overestimated binding energies. Also, we systematically studied the electronic properties by monitoring how the adsorption of each metal atom modifies the electronic band structure, the conductivity, the charge distribution, and the magnetic properties of the CNT.
J Mol Model (2014) 20:2094
Page 3 of 8, 2094
The interaction between Ni and the CNT modifies the C1– C2 bond length (compared with the atom pair bond length in the pristine CNT), and has the most negative binding energy (−0.935 eV), which indicates the strongest adsorption among the three metals studied. The binding energy of Pb (−0.895 eV) is slightly less negative than that for the CNT+Ni system, and Pb binding results in a slight variation in the C1–C2 bond length. Finally, the adsorption of Cd at the CNT surface has the least negative binding energy (−0.713 eV) and leads to the same change in C1–C2 bond length as Pb adsorption. The nanotube carbon–metal bond lengths for Pb and Cd are similar. The geometric effects of adsorbing these metals at the CNT surface may be explained by noting that Ni has the lowest covalent radius among the metals studied, whereas the covalent radii for Cd and Pb are rather similar [43–45]. In each case, the negative binding energy indicates the exothermic and stable adsorption of the heavy metal at the CNT surface [46]. Therefore, such CNTs can be used as a material for actively adsorbing the heavy metals Ni, Pb, and Cd. However, since Ni and Pb are chemisorbed, the cost of regenerating (and thus reutilizing) the bare CNT may be high, because the chemical bonds between the metal and CNT atoms must be broken to do this. Cd is physisorbed with a lower interaction energy, so it is more easily removed from the surface of the CNT, allowing the CNT to be reutilized.
In the first step, we studied the total energy as a function of the CNT–metal distance. In the second step, in order to study the CNT–metal interaction in depth, full geometric optimization was performed, starting from the minimum-energy configuration generated in the first step. The total energy as a function of the CNT–metal distance is shown in Fig. 1. The total energies for the three complexes are negative, indicating that the complexes formed are chemically stable (the minimum energy and corresponding optimal distance from the metal to the CNT surface are shown in each graph). A relatively short CNT–metal distance was obtained for Ni. Cd and Pb are separated from the CNT surface by the same distance. The shapes of the curves presented in Fig. 1 can be used to estimate the strengths of the interactions between the metals and the CNT. The interaction curve for the CNT+Ni system shows a deep well (strong interaction strength), whereas a shallower well is seen for the CNT+Pb system (medium interaction strength), and the shallowest well is observed for the CNT+Cd system (low interaction strength). The fully relaxed structures are shown in Fig. 2 for the three complexes CNT+Ni, CNT+Cd, and CNT+Pb. This figure suggests that the Ni and Pb are chemisorbed on the CNT surface and bond directly with two carbon atoms (labeled C1 and C2). The Cd atom does not bond; it is physisorbed at the CNT surface. One of the primary differences between chemisorption and physisorption is the resulting perturbations of the electronic states of the adsorbent and adsorbate [42]. Later (in the next section), we will see that the calculated electronic properties of the studied systems are modified upon adsorption, which confirms this phenomenon. The bond lengths (measured from the relaxed structure obtained following geometry optimization) and binding energies (calculated using Eq. 1) are shown in Table 1. Fig. 1 Total energy vs. CNT– metal distance for a CNT+Ni, b CNT+Cd, and c CNT+Pb. The values for the minimum energy (E0) and the corresponding optimal distance (r0) are shown for each system
Electronic properties The electronic properties shown in this study were calculated using the parameters described in “Modeling details.” Figure 3 shows the electronic band structure for the pristine CNT (Fig. 3a), the CNT+Ni (Fig. 3b), the CNT+Cd (Fig. 3c), and the CNT+Pb (Fig. 3d) systems. The electronic bands
Energy (eV) 1
2
(c) CNT+Pb
(b) CNT+Cd
(a) CNT+Ni
r 0 =1.64 Å
r 0 =2.69 Å
r 0 =2.69 Å
E0=-19565eV
E0= -19846eV
E0= -20671eV
3
4
1
2
3
Distance CNT - metal (Å)
4
1
2
3
4
2094, Page 4 of 8
J Mol Model (2014) 20:2094
Fig. 2 Structural configurations for the lowest-energy complexes: a CNT+Ni, b CNT+Cd, and c CNT+Pb (picture rendered using XCrySDen software [41])
calculated for the CNT+Cd system are similar to the bands calculated for the pristine CNT. This finding is consistent with our assumption that the Cd is physisorbed; therefore, few changes are induced in the electronic properties of the system in this case. For Ni and Pb, the modifications to the electronic structure of the CNT are more pronounced, which is consistent with chemisorption. Nevertheless, in each case, metal adsorption induces a reduction/change in the electronic gap. Adsorption of Ni on the CNT triggers electronic band separation of the spin-up/down electrons. The corresponding bands are shown in Fig. 4b as black solid lines and red dashdot lines for the spin-up and spin-down electrons, respectively (for comparison, the band structure for the pristine CNT is shown in Fig. 4a). This separation translates into an electronic spin polarization of 0.44 μB (where μB is a Bohr magneton). Therefore, Ni adsorption alters the magnetic properties of the CNT. Ni adsorption at the CNT surface increases the number of bands—more so for the valence band region than for the conduction band region. Notably, minor polarization is observed near the Fermi energy, whereas major polarization occurs in the valence band. Even with this small polarization in the conduction band, a device constructed with a CNT in the active region could be biased such that spin-polarized currents are transported [24]. An additional modification is a change in the electronic character of the nanotube. Before Ni adsorption, the CNT is a semiconductor with a band gap of approximately 0.946 eV. After adsorption, its character changes: it becomes a metallic conductor. Ni adsorption also results in the transfer of 0.240e of charge to the nanotube, which indicates that Ni acts as a donor. Figure 4 shows the total density of states (DOS) and the partial density of states (PDOS) for the CNT+Ni system (both Table 1 Length of the C1–C2 bond (dC1−C2), length of the metal–carbon bond (dC−M), and the fully BSSE-corrected binding energy (Eb) for the isolated CNT and for various metal–CNT systems System
dC1−C2 (Å)
dC−M (Å)
Eb (eV)
CNT CNT+Ni CNT+Cd CNT+Pb
1.42 1.46 1.44 1.44
– 1.83 2.63 2.54
– −0.935 −0.713 −0.895
given in arbitrary units). In parts c–e, the left side of each graph corresponds to the spin-down electronic contribution (denoted by ↓), whereas the right side corresponds to the spinup contribution (denoted by ↑). The Fermi energy (EF) for each system is shown as a green dotted line. The PDOS calculations (Fig. 4d and e) show that the appearance of localized states near the Fermi energy and −3.5 eV is primarily due to the hybridization of the empty 2p carbon orbitals with the 3d orbitals from Ni. This hybridization near the Fermi energy completes the bond states and leads to chemical bond formation. Figure 4b shows that spin polarization increases as the energy decreases. Figure 4e demonstrates that this behavior is associated with a contribution from the electrons, principally those in the 4s and 3d orbitals of Ni. The PDOS calculation shows that the 3d orbitals from Ni are responsible for the increase in the number of electronic bands. Figure 5 shows the electronic bands, the DOS, and the PDOS of the system after Cd adsorption. This figure shows that the bands (Fig. 5b) are only slightly altered from those of the pristine CNT system (Fig. 5a). Among the visible modifications is a decrease in the band gap to 0.725 eV (so the system retains its semiconducting character), the appearance of a band near −4.0 eV, and null spin polarization. Due to Cd adsorption at the nanotube surface, there is a slight transfer of charge to the nanotube (0.034e) and a small increase in the Fermi energy. A density of states analysis shows that there are no available states near the Fermi energy. Conversely, a new band appears near −4.0 eV. This band is associated with the hybridization of the 5s and 2p orbitals from the Cd and the C1 and C2 carbons, respectively, with the largest contribution coming from the Cd 5s orbital. As this alignment is far from the Fermi energy (EF, green dotted lines in Fig. 5), a chemical bond is not formed. In parts c–e for Fig. 5, the left side of each graph corresponds to the spin-down electronic contribution (denoted by ↓), whereas the right side corresponds to the spin-up contribution (denoted by ↑). In the three graphs, the curves for both electron types are entirely symmetric, which confirms that there is null spin polarization for the CNT+Cd system. The few modifications to the electronic band structure and the density of states for the CNT+Cd system confirm that physisorption occurs, not chemisorption.
J Mol Model (2014) 20:2094
Page 5 of 8, 2094
Fig. 3 Calculated electronic band structures for a pristine CNT, b CNT+Ni, c CNT+Cd, and d CNT+Pb. The green dashed line corresponds to the Fermi energy (EF), the black solid lines indicate bands for spin-down electrons, and the red dash-dot lines represent bands for spin-up electrons
-2.5
(a) CNT
(b) CNT+Ni
(c) CNT+Cd
(d) CNT+Pb
-3.0
Energy (eV)
-3.5
-4.0
-4.5
-5.0
Z
Z
Z
Z
EF
The electronic band structure as well as the DOS and the PDOS for the system after Pb adsorption are shown in Fig. 6. The band structure (Fig. 6a for the pristine CNT and Fig. 6b for CNT+Pb) indicates that Pb adsorption on the CNT triggers pronounced alterations in the bands, primarily near the Fermi energy (EF, green dotted lines), indicating an increase in CNT
Fig. 4 Electronic band structures for a pristine CNT and b CNT+Ni systems; c total density of states for the CNT+Ni system; d partial density of states for the two carbon atoms C1 and C2 (the nearest neighbors to the metal); and e the partial density of states for Ni. See text for more details
-2.5
(a)
reactivity. Pb adsorption also induces spin polarization in the system and a change in the electronic character of the nanotube to half-metallic. Polarization (translated into band splitting of the spin-up and spin-down electronic configurations) is concentrated in the conduction band. As the conduction band is responsible for electronic transport, the induced splitting
(b)
(c)
(d)
(e)
-3.0
Energy (eV)
-3.5
-4.0
-4.5
-5.0 Z
Z -28
0
DOS EF
28 -0.1
0.0
2p C1 2p C2
0.1 -22
0
22
3d Ni 4s Ni
2094, Page 6 of 8
J Mol Model (2014) 20:2094
Fig. 5 Electronic band structures for a pristine CNT and b CNT+ Cd systems, c total density of states for the CNT+Cd system, d partial density of states of the two carbon atoms C1 and C2 (the nearest neighbors to the metal); and e the partial density of states for Cd. See text for more details
-2.5
(b)
(a)
(c)
(e)
(d)
-3.0
Energy (eV)
-3.5
-4.0
-4.5
-5.0 Z
Z -21
0
21 -1
DOS
Fig. 6 Electronic band structures of a pristine CNT and b CNT+Pb systems; c total density of states of the CNT+Pb system; d partial density of states of the two carbon atoms C1 and C2 (the nearest neighbors to the metal); and e the partial density of states of Pb. See text for more details
(a)
1 -13
0
13
4d Cd 5s Cd
2p C1 2p C2
EF
could be utilized in spin transport devices [24]. The induced spin polarization is on the order of 1.75 μB, leading to new magnetic properties for the nanotube. The magnetism induced by Pb (an sp-like metal) is due to the mobility of the four electrons from the 6s and 6p orbitals and the influence of the 5d orbital. In previous works [47, 48], the authors have shown (experimentally and theoretically) that various conducting channels occur in atomic lead. Based on those works, it seems that three wide-open transmission
0
channels of spz, py, and py character appear. The effect associated with the d orbitals is to move the p bands to higher energies. The appearance of these channels permits the existence of unpaired electrons, and thus the possibility of inducing magnetism in the system. Figure 6d and e show the hybridization of the 2p orbitals from carbon atoms C1 and C2 with the 6p orbital from Pb. As this hybridization occurs near to the Fermi energy, chemical bond formation between the Pb and C1, as well as C2, is
(b)
(d)
(c)
(e)
-2.5
-3.0
Energy (eV)
-3.5
-4.0
-4.5
-5.0 Z
Z -45
0
DOS EF
45 -1
0
2p C1 2p C2
1 -26
0
5d Pb 6p Pb
26
J Mol Model (2014) 20:2094
favored. This hybridization results in a change transfer of 0.018e from the CNT to the metal. Further, a comparison of Fig. 6b, c, and e shows that the 6p orbital from Pb may be responsible for the intense polarization in the conduction band and Fermi energy region.
Page 7 of 8, 2094 Estudos e Projetos-Ministério de Ciência e Tecnologia (FINEP-MCT) grant from Centro Nacional de Processamento de Alto Desempenho em São Paulo (CENAPAD-SP).
References Conclusions Using first-principles calculations, we characterized the adsorption of Ni, Cd, and Pb at the surface of a CNT. Based on our adsorption studies, we can conclude that the three metals are stably adsorbed at the nanotube surface. Therefore, CNTs are a feasible active material for filters that retain such metals. If we only consider the binding energies (even when the binding energy for Pb is 25% lower than that for Cd adsorption), we cannot establish whether the type of adsorption that occurs is physisorption or chemisorption. One way to resolve this issue is to use, for example, the quantum theory of atoms in molecules (QTAIM) [49] and the Bader charge partition scheme [50–52]. Using the Bader theory, the types of bonds that occur can be determined from the topological properties of the electron density. However, we did not perform such a study in the present work. Instead, we followed the recommendation that physisorption does not involve a significant change in the electronic orbital patterns of the systems involved [53]. From the calculated electronic band structures (Figs. 4, 5, and 6), we can see that major modifications occur to the band structure when Ni and Pb atoms are adsorbed, whereas only minor modifications occur when Cd is adsorbed. Thus, Cd is physisorbed while Ni and Pb are chemisorbed. This is confirmed by a comparison of the electronic properties of the nanotube before and after metal adsorption in each case. As Cd is physisorbed, the binding strength is low, which facilitates inexpensive nanotube reutilization. For Ni and Pb, the binding strength is higher, which indicates that nanotube reutilization may be a difficult task. The electronic and magnetic properties of the CNT are affected by metal adsorption. Consequently, such variations can be explored for use as the primary properties employed in devices for sensing heavy metals in solutions. Such a device could be biased such that the adsorption of a particular metal produces a current of a particular intensity [6, 19]. The spin polarization induced by Ni and Pb adsorption could be used not only for detection but also as the primary characteristic of a nano-spintronic device that uses spinpolarized currents. Acknowledgments We would like to acknowledge financial support from the Brazilian agencies Conselho Nacional de Pesquisa (CNPq), Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), and Fundação de Amparo à Pesquisa do Estado de Minas Gerais (FAPEMIG). Some of the results presented here were developed using a Universidade de Campinas (UNICAMP) / Financiadora de
1. Ives AR, Cardinale BJ (2004) Nature 429:174 2. Chen T, Liu X, Zhu M, Zhao K, Wu J, Xu J, Huang P (2008) Environ Pollut 151:67 3. Duran A, Tuzen M, Soylak M (2009) J Hazard Mater 169:466 4. Kong J, Franklin NR, Zhou C, Chapline MG, Peng S, Cho K, Dai H (2000) Science 287(5453):622 5. Collins PG, Bradley K, Ishigami M, Zettl A (2000) Science 287(5459):1801 6. Kauffman D, Star A (2008) Angew Chem Int Ed 47:6550 7. Dresselhaus M, Dresselhaus G, Jorio A (2004) Annu Rev Mate Res 34:247 8. Snow E, Perkins F, Robinson J (2006) Chem Soc Rev 35:790 9. Ravelo-Pérez LM, Herrera-Herrera AV, Hernández-Borges J, Ángel Rodríguez-Delgado M (2010) J Chromatogr A 1217:2618 10. Iijima S (1991) Nature 354:56 11. Hayes KE, Lee HS (2012) Chem Phys 393:96 12. Park M, Kim BH, Kim S, Han DS, Kim G, Lee KR (2011) Carbon 49:811 13. Zhang ZW, Zheng WT, Jiang Q (2011) Phys Chem Chem Phys 13: 9483 14. Mota R, Fagan SB, Fazzio A (2007) Surf Sci 601:4102 15. Li K, Wang W, Cao D (2011) Sensors Actuators B 159:171 16. Zhao JX, Ding YH (2008) Mater Chem Phys 110:411 17. Shalabi A, Abdel Aal S, Assem M, Abdel Halim WS (2013) Int J Hydrog Energy 38:140 18. Girão EC, Liebold-Ribeiro Y, Batista JA, Barros EB, Fagan SB, Filho JM, Dresselhaus MS, Filho AGS (2010) Phys Chem Chem Phys 12: 1518 19. Allen B, Kichambare P, Star A (2007) Adv Mater 19(11):1439 20. Cadore AR, Zanella I, de Menezes VM, Rossato J, Mota R, Fagan SB (2012) Phys Chem Chem Phys 14:16737 21. Veloso M, Filho AS, Filho JM, Fagan S, Mota R (2006) Chem Phys Lett 430:71 22. Ghaedi M, Montazerzohori M, Rahimi N, Biysreh MN (2013) J Ind Eng Chem 19(5):1477 23. Zhang Y, Franklin N, Chen R, Dai H (2000) Chem Phys Lett 331:35 24. Flatté M (2007) IEEE Trans Electron Devices 54(5):907 25. Fürst JA, Brandbyge M, Jauho AP, Stokbro K (2008) Phys Rev 78: 195405 26. Blase X, Margine E (2009) Appl Phys Lett 94:173103 27. Zhang G, Liu X, Wang C, Yao Y, Zhang J, Ho K (2013) J Phys Condens Matter 25:105302 28. Serp P, Corrias M, Kalck P (2003) Appl Catal A 28:337 29. Andriotis AN, Menon M, Froudakis G (2000) Phys Rev Lett 85:3193 30. Soler J, Artacho E, Gale J, García A, Junquera J, Ordejón P, SánchezPortal D (2002) J Phys Condens Matter 14:2745 31. Ceperley DM, Alder BJ (1980) Phys Rev Lett 45:566 32. Perdew JP, Zunger A (1981) Phys Rev B 23:5048 33. Troullier N, Martins JL (1991) Phys Rev B 43:1993 34. Artacho E, Sánchez-Portal D, Ordejón P, García A, Soler J (1999) Phys Stat Sol B 215:809 35. Monkhorst H, Pack J (1976) Phys Rev B 13(12):5188 36. Boys S, Bernardi F (1970) Mol Phys 19:553 37. Mulliken R (1955) J Chem Phys 23(10):1833 38. Mulliken R (1955) J Chem Phys 23(10):1841 39. Durgun E, Dag S, Bagci V, Gülseren O, Yildirim T, Ciraci S (2003) Phys Rev B 67:201401
2094, Page 8 of 8 40. Durgun E, Dag S, Ciraci S, Gülseren O (2004) J Phys Chem B 108:575 41. Kokalj A (2003) Comput Mater Sci 28(2):155 42. Everett D (2001) Manual on definitions, terminology and symbols in colloid and surface chemistry (internet version). Division of Physical Chemistry, IUPAC, Zürich 43. Cordero B, Gomez V, Platero-Prats AE, Reves M, Echeverria J, Cremades E, Barragan F, Alvarez S (2008) Dalton Trans 21:2832 44. Pyykkö P, Atsumi M (2009) Chem Eur J 15(1):186 45. Pyykkö P, Atsumi M (2009) Chem Eur J 15(46):12770 46. Xie Y, Huo YP, Zhang JM (2012) Appl Surf Sci 258:6391 47. Cuevas J, Levy Yeyati A, Martín-Rodero A (1998) Phys Rev Lett 80(5):1066
J Mol Model (2014) 20:2094 48. Cuevas J, Levy Yeyati A, Martín-Rodero A, Rubio Bollinger G, Untiedt C, Agraït N (1998) Phys Rev Lett 81(14):2990 49. Bader RFW (1994) Atoms in molecules: a quantum theory. Clarendon, Oxford 50. Henkelman G, Arnaldsson A, Jónsson H (2006) Comput Mater Sci 36:354 51. Sanville E, Kenny S, Smith R, Henkelman G (2007) J Comp Chem 28(5):899 52. Tang W, Sanville E, Henkelman G (2009) J Phys Condens Matter 21: 084204 53. IUPAC (1997) Compendium of chemical terminology. Gold book, 2nd edn. Blackwell, Oxford. doi:10.1351/goldbook, V. 2.3.2
ORIGINAL PAPER
First-principles calculations of nickel, cadmium, and lead adsorption on a single-walled (10,0) carbon nanotube Mirele Bastos & Ihosvany Camps
Received: 11 August 2013 / Accepted: 25 November 2013 / Published online: 11 February 2014 # Springer-Verlag Berlin Heidelberg 2014
Abstract The adsorption of Ni, Cd, and Pb on a zigzag (10, 0) carbon nanotube (CNT) surface was investigated using density functional theory. Binding energy calculations were performed, and the results indicated that the three metals are stably adsorbed on the nanotube surface. Moreover, the results showed that Cd is physisorbed whereas Ni and Pb are chemisorbed. Our studies show that the electronic properties of the CNT are modified by the chemisorption mechanism (Ni and Pb). After Ni and Pb adsorption, the nanotube changes from being a semiconductor to a metallic conductor. The nanotube remains semiconductive upon Cd physisorption, although a decrease in the band gap is observed. Also, Ni or Pb adsorption triggers a change in the magnetism of the nanotube through the induction of spin polarization. Not only can these results of our calculations be used to explain the adsorption mechanisms of these heavy metals on the CNT, but they are also useful for evaluating the potential of carbon nanotubes (CNTs) to act as filters and sensors of such metals. Keywords Nanostructures . Ab initio calculations . Electronic structure . Band structure
Introduction In recent decades, technological and industrial advances and a lack of care in toxic material disposal have resulted in the gradual extermination of many plant and animal species [1]. The pollutants that cause the most concern include heavy metals, such as Ni, Cd, Pb, Cr, Cu, Hg, Zn, and Mn. There M. Bastos : I. Camps (*) Laboratório de Modelagem Computacional—LaModel, Instituto de Ciências Exatas, Universidade Federal de Alfenas—Unifal-MG, CEP 37130-000 Alfenas, Minas Gerais, Brazil e-mail: [email protected]
is great apprehension over the exposure of populations to heavy metals because they are not biodegradable and they have long biological lifetimes, which hinder their removal from the body [2]. One way to circumvent this problem is to use solid-phase materials that retain heavy metals and are simple and inexpensive. Such a material should have a large surface area, a small size, and high chemical stability. Among the materials that exhibit such characteristics are carbon nanotubes (CNTs) [3–10]. A search for new materials with high adsorption capacities and surface areas suggests that CNTs are strong candidates. Many theoretical and experimental studies have attempted to unravel the interaction mechanisms of metals [11, 12], gases [6, 13–17], and organic molecules [18, 19] with pure CNTs (with various geometries), CNTs doped with additional elements [14–17, 20], and CNTs decorated with organic functional groups [12, 18, 21, 22]. Experimental studies on singlewalled carbon nanotube field-effect transistors have already been reviewed [6, 19]. However, CNTs that interact with metals, especially transition metals, have been studied for other applications too, such as in nanodevices [15, 23], spintronics [24–27], and metal-catalytic systems [28, 29]. In the work described in the present paper, we studied the interactions between the heavy metals Ni, Cd, and Pb and a single-walled CNT. We chose these metals due to the growth of industrial activities such as the burning of coal and oil (for power generation), as these activities are contributing to the release of Ni, Cd, and Pb into the environment. Furthermore, there is no comprehensive study in the literature on the interaction of these metals with a single-walled CNT. We used density functional theory to calculate the differences in the electronic properties of the CNT before and after metal adsorption. This paper is organized as follows. In the next section (“Modeling details”), we describe the methodology and parameters used for the calculations and the system configuration employed. Subsequent sections discuss results relating to
2094, Page 2 of 8
J Mol Model (2014) 20:2094
adsorption (“Ni, Cd, and Pb adsorption at the CNT surface”) and the electronic properties of the nanotube (“Electronic properties”). In the final section of the paper (“Conclusions”), we draw conclusions from the results we obtained.
Modeling details We used density functional theory (DFT) as implemented in the SIESTA code [30] and the local density approximation (LDA) for the exchange-correlation potential. Both the Ceperley–Alder (CA) [31] scheme and the Perdew and Zunger parameters [32] (PZ) were used. To avoid the need to explicitly treat the core electrons, the norm-conserving Troullier–Martins pseudopotentials [33] were used. For the valence electrons, we used a split-valence double-zeta basis set with polarization functions (DZP) [34]. Periodic boundary conditions were applied using the supercell approach. A (10, 0) zigzag pure CNT with 120 atoms, or the same nanotube plus a heavy metal (121 atoms), was placed (oriented in the z-direction) inside a 38×25×12 Å cell. Using this configuration, each system image was separated by at least ∼10 and ∼22 Å in the x- and y-directions, respectively, to avoid interactions between two neighboring systems. To produce sufficiently accurate results, convergence studies were performed for the grid (mesh) cutoff energy and the number of k-points. Total energy convergence was possible for the systems tested here using a mesh cutoff of ∼200 Ry; however, we used 300 Ry to ensure accuracy. To estimate the required number of k-points, the Brillouin zone was sampled in accordance with Monkhorst and Pack [35]; convergence was obtained for a 1×1×8 set, but a 1×1×11 k-point sampling set was actually used in this work. The electronic properties were calculated for systems with fully relaxed atomic coordinates obtained by performing geometry optimization until the Hellmann–Feynman forces were below 0.03 eV Å −1. For the adsorption studies, the CNT+metal system energy was calculated starting with a manually specified metal position (supplied by an input file). The binding energy (Eb) for each heavy metal adsorbed on the CNT was calculated using the following expression [36]: E b ¼ E CNTþM −ECNT −EM −E CC ;
ð1Þ
where ECNT and EM are the Kohn–Sham energies for the CNT and metal, respectively, ECNT+M is the Kohn–Sham energy of the complex CNT+metal, and ECC is the counterpoise correction that accounts for the effects of the basis set superposition error (BSSE), as follows: E CC ¼ E ghost−M þ Eghost−CNT ;
ð2Þ
where Eghost−M and Eghost−CNT are the metal and CNT Kohn– Sham energies, respectively, and the atoms were replaced by ghost atoms [30]. The atomic charge distribution can be calculated from first principles. In this work, we examined the topological charge density distribution using the Mulliken population analysis [37, 38] implemented in the SIESTA code. Variations in the charge on the metal due to transfer between the CNT and the metal were calculated using the following expression: iso ΔQM ¼ Qabs M −QM ;
ð3Þ
where Qabs M is the total charge on the metal after adsorption and Qiso M is the total charge on the isolated metal.
Results Ni, Cd, and Pb adsorption at the CNT surface In the work reported here, we were interested in the interaction between a heavy metal (either Ni, Cd, or Pb) and a (10, 0) zigzag CNT. The CNT surface has different areas where the metal can interact. The physical origin of these different adsorption sites is mainly the shape of the overlap between the frontier atomic orbitals of carbon atoms (from the CNT) at each site and the frontier orbitals of the adsorbed atoms. Other factors that can affect the adsorption include the electronic distribution and charge polarization effects on the CNT sites and on the added metal. It is possible to calculate and plot the frontier orbitals in those regions. Following the results of Durgun et al. [39, 40], we only studied the interaction of each metal with the most probable binding site: the bridge site. In these works, the authors mainly calculated the binding energies of several metals adsorbed on a CNT. In terms of the electronic properties, they only showed the electronic bands after the adsorption of an Al, C, Si, or Ti atom (none of which are studied in the present work). Thus, in contrast to the works of Durgun et al. [39, 40], the present work represents a more detailed study of the adsorption of Ni, Cd, and Pb on a CNT, including a deep exploration of the mechanisms involved and the resulting modifications to the electronic structure of the CNT. To accomplish this, the binding energies were calculated using a more complete theory: the counterpoise correction, which accounts for the effects of the basis set superposition error (BSSE). This correction is used to avoid under/overestimated binding energies. Also, we systematically studied the electronic properties by monitoring how the adsorption of each metal atom modifies the electronic band structure, the conductivity, the charge distribution, and the magnetic properties of the CNT.
J Mol Model (2014) 20:2094
Page 3 of 8, 2094
The interaction between Ni and the CNT modifies the C1– C2 bond length (compared with the atom pair bond length in the pristine CNT), and has the most negative binding energy (−0.935 eV), which indicates the strongest adsorption among the three metals studied. The binding energy of Pb (−0.895 eV) is slightly less negative than that for the CNT+Ni system, and Pb binding results in a slight variation in the C1–C2 bond length. Finally, the adsorption of Cd at the CNT surface has the least negative binding energy (−0.713 eV) and leads to the same change in C1–C2 bond length as Pb adsorption. The nanotube carbon–metal bond lengths for Pb and Cd are similar. The geometric effects of adsorbing these metals at the CNT surface may be explained by noting that Ni has the lowest covalent radius among the metals studied, whereas the covalent radii for Cd and Pb are rather similar [43–45]. In each case, the negative binding energy indicates the exothermic and stable adsorption of the heavy metal at the CNT surface [46]. Therefore, such CNTs can be used as a material for actively adsorbing the heavy metals Ni, Pb, and Cd. However, since Ni and Pb are chemisorbed, the cost of regenerating (and thus reutilizing) the bare CNT may be high, because the chemical bonds between the metal and CNT atoms must be broken to do this. Cd is physisorbed with a lower interaction energy, so it is more easily removed from the surface of the CNT, allowing the CNT to be reutilized.
In the first step, we studied the total energy as a function of the CNT–metal distance. In the second step, in order to study the CNT–metal interaction in depth, full geometric optimization was performed, starting from the minimum-energy configuration generated in the first step. The total energy as a function of the CNT–metal distance is shown in Fig. 1. The total energies for the three complexes are negative, indicating that the complexes formed are chemically stable (the minimum energy and corresponding optimal distance from the metal to the CNT surface are shown in each graph). A relatively short CNT–metal distance was obtained for Ni. Cd and Pb are separated from the CNT surface by the same distance. The shapes of the curves presented in Fig. 1 can be used to estimate the strengths of the interactions between the metals and the CNT. The interaction curve for the CNT+Ni system shows a deep well (strong interaction strength), whereas a shallower well is seen for the CNT+Pb system (medium interaction strength), and the shallowest well is observed for the CNT+Cd system (low interaction strength). The fully relaxed structures are shown in Fig. 2 for the three complexes CNT+Ni, CNT+Cd, and CNT+Pb. This figure suggests that the Ni and Pb are chemisorbed on the CNT surface and bond directly with two carbon atoms (labeled C1 and C2). The Cd atom does not bond; it is physisorbed at the CNT surface. One of the primary differences between chemisorption and physisorption is the resulting perturbations of the electronic states of the adsorbent and adsorbate [42]. Later (in the next section), we will see that the calculated electronic properties of the studied systems are modified upon adsorption, which confirms this phenomenon. The bond lengths (measured from the relaxed structure obtained following geometry optimization) and binding energies (calculated using Eq. 1) are shown in Table 1. Fig. 1 Total energy vs. CNT– metal distance for a CNT+Ni, b CNT+Cd, and c CNT+Pb. The values for the minimum energy (E0) and the corresponding optimal distance (r0) are shown for each system
Electronic properties The electronic properties shown in this study were calculated using the parameters described in “Modeling details.” Figure 3 shows the electronic band structure for the pristine CNT (Fig. 3a), the CNT+Ni (Fig. 3b), the CNT+Cd (Fig. 3c), and the CNT+Pb (Fig. 3d) systems. The electronic bands
Energy (eV) 1
2
(c) CNT+Pb
(b) CNT+Cd
(a) CNT+Ni
r 0 =1.64 Å
r 0 =2.69 Å
r 0 =2.69 Å
E0=-19565eV
E0= -19846eV
E0= -20671eV
3
4
1
2
3
Distance CNT - metal (Å)
4
1
2
3
4
2094, Page 4 of 8
J Mol Model (2014) 20:2094
Fig. 2 Structural configurations for the lowest-energy complexes: a CNT+Ni, b CNT+Cd, and c CNT+Pb (picture rendered using XCrySDen software [41])
calculated for the CNT+Cd system are similar to the bands calculated for the pristine CNT. This finding is consistent with our assumption that the Cd is physisorbed; therefore, few changes are induced in the electronic properties of the system in this case. For Ni and Pb, the modifications to the electronic structure of the CNT are more pronounced, which is consistent with chemisorption. Nevertheless, in each case, metal adsorption induces a reduction/change in the electronic gap. Adsorption of Ni on the CNT triggers electronic band separation of the spin-up/down electrons. The corresponding bands are shown in Fig. 4b as black solid lines and red dashdot lines for the spin-up and spin-down electrons, respectively (for comparison, the band structure for the pristine CNT is shown in Fig. 4a). This separation translates into an electronic spin polarization of 0.44 μB (where μB is a Bohr magneton). Therefore, Ni adsorption alters the magnetic properties of the CNT. Ni adsorption at the CNT surface increases the number of bands—more so for the valence band region than for the conduction band region. Notably, minor polarization is observed near the Fermi energy, whereas major polarization occurs in the valence band. Even with this small polarization in the conduction band, a device constructed with a CNT in the active region could be biased such that spin-polarized currents are transported [24]. An additional modification is a change in the electronic character of the nanotube. Before Ni adsorption, the CNT is a semiconductor with a band gap of approximately 0.946 eV. After adsorption, its character changes: it becomes a metallic conductor. Ni adsorption also results in the transfer of 0.240e of charge to the nanotube, which indicates that Ni acts as a donor. Figure 4 shows the total density of states (DOS) and the partial density of states (PDOS) for the CNT+Ni system (both Table 1 Length of the C1–C2 bond (dC1−C2), length of the metal–carbon bond (dC−M), and the fully BSSE-corrected binding energy (Eb) for the isolated CNT and for various metal–CNT systems System
dC1−C2 (Å)
dC−M (Å)
Eb (eV)
CNT CNT+Ni CNT+Cd CNT+Pb
1.42 1.46 1.44 1.44
– 1.83 2.63 2.54
– −0.935 −0.713 −0.895
given in arbitrary units). In parts c–e, the left side of each graph corresponds to the spin-down electronic contribution (denoted by ↓), whereas the right side corresponds to the spinup contribution (denoted by ↑). The Fermi energy (EF) for each system is shown as a green dotted line. The PDOS calculations (Fig. 4d and e) show that the appearance of localized states near the Fermi energy and −3.5 eV is primarily due to the hybridization of the empty 2p carbon orbitals with the 3d orbitals from Ni. This hybridization near the Fermi energy completes the bond states and leads to chemical bond formation. Figure 4b shows that spin polarization increases as the energy decreases. Figure 4e demonstrates that this behavior is associated with a contribution from the electrons, principally those in the 4s and 3d orbitals of Ni. The PDOS calculation shows that the 3d orbitals from Ni are responsible for the increase in the number of electronic bands. Figure 5 shows the electronic bands, the DOS, and the PDOS of the system after Cd adsorption. This figure shows that the bands (Fig. 5b) are only slightly altered from those of the pristine CNT system (Fig. 5a). Among the visible modifications is a decrease in the band gap to 0.725 eV (so the system retains its semiconducting character), the appearance of a band near −4.0 eV, and null spin polarization. Due to Cd adsorption at the nanotube surface, there is a slight transfer of charge to the nanotube (0.034e) and a small increase in the Fermi energy. A density of states analysis shows that there are no available states near the Fermi energy. Conversely, a new band appears near −4.0 eV. This band is associated with the hybridization of the 5s and 2p orbitals from the Cd and the C1 and C2 carbons, respectively, with the largest contribution coming from the Cd 5s orbital. As this alignment is far from the Fermi energy (EF, green dotted lines in Fig. 5), a chemical bond is not formed. In parts c–e for Fig. 5, the left side of each graph corresponds to the spin-down electronic contribution (denoted by ↓), whereas the right side corresponds to the spin-up contribution (denoted by ↑). In the three graphs, the curves for both electron types are entirely symmetric, which confirms that there is null spin polarization for the CNT+Cd system. The few modifications to the electronic band structure and the density of states for the CNT+Cd system confirm that physisorption occurs, not chemisorption.
J Mol Model (2014) 20:2094
Page 5 of 8, 2094
Fig. 3 Calculated electronic band structures for a pristine CNT, b CNT+Ni, c CNT+Cd, and d CNT+Pb. The green dashed line corresponds to the Fermi energy (EF), the black solid lines indicate bands for spin-down electrons, and the red dash-dot lines represent bands for spin-up electrons
-2.5
(a) CNT
(b) CNT+Ni
(c) CNT+Cd
(d) CNT+Pb
-3.0
Energy (eV)
-3.5
-4.0
-4.5
-5.0
Z
Z
Z
Z
EF
The electronic band structure as well as the DOS and the PDOS for the system after Pb adsorption are shown in Fig. 6. The band structure (Fig. 6a for the pristine CNT and Fig. 6b for CNT+Pb) indicates that Pb adsorption on the CNT triggers pronounced alterations in the bands, primarily near the Fermi energy (EF, green dotted lines), indicating an increase in CNT
Fig. 4 Electronic band structures for a pristine CNT and b CNT+Ni systems; c total density of states for the CNT+Ni system; d partial density of states for the two carbon atoms C1 and C2 (the nearest neighbors to the metal); and e the partial density of states for Ni. See text for more details
-2.5
(a)
reactivity. Pb adsorption also induces spin polarization in the system and a change in the electronic character of the nanotube to half-metallic. Polarization (translated into band splitting of the spin-up and spin-down electronic configurations) is concentrated in the conduction band. As the conduction band is responsible for electronic transport, the induced splitting
(b)
(c)
(d)
(e)
-3.0
Energy (eV)
-3.5
-4.0
-4.5
-5.0 Z
Z -28
0
DOS EF
28 -0.1
0.0
2p C1 2p C2
0.1 -22
0
22
3d Ni 4s Ni
2094, Page 6 of 8
J Mol Model (2014) 20:2094
Fig. 5 Electronic band structures for a pristine CNT and b CNT+ Cd systems, c total density of states for the CNT+Cd system, d partial density of states of the two carbon atoms C1 and C2 (the nearest neighbors to the metal); and e the partial density of states for Cd. See text for more details
-2.5
(b)
(a)
(c)
(e)
(d)
-3.0
Energy (eV)
-3.5
-4.0
-4.5
-5.0 Z
Z -21
0
21 -1
DOS
Fig. 6 Electronic band structures of a pristine CNT and b CNT+Pb systems; c total density of states of the CNT+Pb system; d partial density of states of the two carbon atoms C1 and C2 (the nearest neighbors to the metal); and e the partial density of states of Pb. See text for more details
(a)
1 -13
0
13
4d Cd 5s Cd
2p C1 2p C2
EF
could be utilized in spin transport devices [24]. The induced spin polarization is on the order of 1.75 μB, leading to new magnetic properties for the nanotube. The magnetism induced by Pb (an sp-like metal) is due to the mobility of the four electrons from the 6s and 6p orbitals and the influence of the 5d orbital. In previous works [47, 48], the authors have shown (experimentally and theoretically) that various conducting channels occur in atomic lead. Based on those works, it seems that three wide-open transmission
0
channels of spz, py, and py character appear. The effect associated with the d orbitals is to move the p bands to higher energies. The appearance of these channels permits the existence of unpaired electrons, and thus the possibility of inducing magnetism in the system. Figure 6d and e show the hybridization of the 2p orbitals from carbon atoms C1 and C2 with the 6p orbital from Pb. As this hybridization occurs near to the Fermi energy, chemical bond formation between the Pb and C1, as well as C2, is
(b)
(d)
(c)
(e)
-2.5
-3.0
Energy (eV)
-3.5
-4.0
-4.5
-5.0 Z
Z -45
0
DOS EF
45 -1
0
2p C1 2p C2
1 -26
0
5d Pb 6p Pb
26
J Mol Model (2014) 20:2094
favored. This hybridization results in a change transfer of 0.018e from the CNT to the metal. Further, a comparison of Fig. 6b, c, and e shows that the 6p orbital from Pb may be responsible for the intense polarization in the conduction band and Fermi energy region.
Page 7 of 8, 2094 Estudos e Projetos-Ministério de Ciência e Tecnologia (FINEP-MCT) grant from Centro Nacional de Processamento de Alto Desempenho em São Paulo (CENAPAD-SP).
References Conclusions Using first-principles calculations, we characterized the adsorption of Ni, Cd, and Pb at the surface of a CNT. Based on our adsorption studies, we can conclude that the three metals are stably adsorbed at the nanotube surface. Therefore, CNTs are a feasible active material for filters that retain such metals. If we only consider the binding energies (even when the binding energy for Pb is 25% lower than that for Cd adsorption), we cannot establish whether the type of adsorption that occurs is physisorption or chemisorption. One way to resolve this issue is to use, for example, the quantum theory of atoms in molecules (QTAIM) [49] and the Bader charge partition scheme [50–52]. Using the Bader theory, the types of bonds that occur can be determined from the topological properties of the electron density. However, we did not perform such a study in the present work. Instead, we followed the recommendation that physisorption does not involve a significant change in the electronic orbital patterns of the systems involved [53]. From the calculated electronic band structures (Figs. 4, 5, and 6), we can see that major modifications occur to the band structure when Ni and Pb atoms are adsorbed, whereas only minor modifications occur when Cd is adsorbed. Thus, Cd is physisorbed while Ni and Pb are chemisorbed. This is confirmed by a comparison of the electronic properties of the nanotube before and after metal adsorption in each case. As Cd is physisorbed, the binding strength is low, which facilitates inexpensive nanotube reutilization. For Ni and Pb, the binding strength is higher, which indicates that nanotube reutilization may be a difficult task. The electronic and magnetic properties of the CNT are affected by metal adsorption. Consequently, such variations can be explored for use as the primary properties employed in devices for sensing heavy metals in solutions. Such a device could be biased such that the adsorption of a particular metal produces a current of a particular intensity [6, 19]. The spin polarization induced by Ni and Pb adsorption could be used not only for detection but also as the primary characteristic of a nano-spintronic device that uses spinpolarized currents. Acknowledgments We would like to acknowledge financial support from the Brazilian agencies Conselho Nacional de Pesquisa (CNPq), Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), and Fundação de Amparo à Pesquisa do Estado de Minas Gerais (FAPEMIG). Some of the results presented here were developed using a Universidade de Campinas (UNICAMP) / Financiadora de
1. Ives AR, Cardinale BJ (2004) Nature 429:174 2. Chen T, Liu X, Zhu M, Zhao K, Wu J, Xu J, Huang P (2008) Environ Pollut 151:67 3. Duran A, Tuzen M, Soylak M (2009) J Hazard Mater 169:466 4. Kong J, Franklin NR, Zhou C, Chapline MG, Peng S, Cho K, Dai H (2000) Science 287(5453):622 5. Collins PG, Bradley K, Ishigami M, Zettl A (2000) Science 287(5459):1801 6. Kauffman D, Star A (2008) Angew Chem Int Ed 47:6550 7. Dresselhaus M, Dresselhaus G, Jorio A (2004) Annu Rev Mate Res 34:247 8. Snow E, Perkins F, Robinson J (2006) Chem Soc Rev 35:790 9. Ravelo-Pérez LM, Herrera-Herrera AV, Hernández-Borges J, Ángel Rodríguez-Delgado M (2010) J Chromatogr A 1217:2618 10. Iijima S (1991) Nature 354:56 11. Hayes KE, Lee HS (2012) Chem Phys 393:96 12. Park M, Kim BH, Kim S, Han DS, Kim G, Lee KR (2011) Carbon 49:811 13. Zhang ZW, Zheng WT, Jiang Q (2011) Phys Chem Chem Phys 13: 9483 14. Mota R, Fagan SB, Fazzio A (2007) Surf Sci 601:4102 15. Li K, Wang W, Cao D (2011) Sensors Actuators B 159:171 16. Zhao JX, Ding YH (2008) Mater Chem Phys 110:411 17. Shalabi A, Abdel Aal S, Assem M, Abdel Halim WS (2013) Int J Hydrog Energy 38:140 18. Girão EC, Liebold-Ribeiro Y, Batista JA, Barros EB, Fagan SB, Filho JM, Dresselhaus MS, Filho AGS (2010) Phys Chem Chem Phys 12: 1518 19. Allen B, Kichambare P, Star A (2007) Adv Mater 19(11):1439 20. Cadore AR, Zanella I, de Menezes VM, Rossato J, Mota R, Fagan SB (2012) Phys Chem Chem Phys 14:16737 21. Veloso M, Filho AS, Filho JM, Fagan S, Mota R (2006) Chem Phys Lett 430:71 22. Ghaedi M, Montazerzohori M, Rahimi N, Biysreh MN (2013) J Ind Eng Chem 19(5):1477 23. Zhang Y, Franklin N, Chen R, Dai H (2000) Chem Phys Lett 331:35 24. Flatté M (2007) IEEE Trans Electron Devices 54(5):907 25. Fürst JA, Brandbyge M, Jauho AP, Stokbro K (2008) Phys Rev 78: 195405 26. Blase X, Margine E (2009) Appl Phys Lett 94:173103 27. Zhang G, Liu X, Wang C, Yao Y, Zhang J, Ho K (2013) J Phys Condens Matter 25:105302 28. Serp P, Corrias M, Kalck P (2003) Appl Catal A 28:337 29. Andriotis AN, Menon M, Froudakis G (2000) Phys Rev Lett 85:3193 30. Soler J, Artacho E, Gale J, García A, Junquera J, Ordejón P, SánchezPortal D (2002) J Phys Condens Matter 14:2745 31. Ceperley DM, Alder BJ (1980) Phys Rev Lett 45:566 32. Perdew JP, Zunger A (1981) Phys Rev B 23:5048 33. Troullier N, Martins JL (1991) Phys Rev B 43:1993 34. Artacho E, Sánchez-Portal D, Ordejón P, García A, Soler J (1999) Phys Stat Sol B 215:809 35. Monkhorst H, Pack J (1976) Phys Rev B 13(12):5188 36. Boys S, Bernardi F (1970) Mol Phys 19:553 37. Mulliken R (1955) J Chem Phys 23(10):1833 38. Mulliken R (1955) J Chem Phys 23(10):1841 39. Durgun E, Dag S, Bagci V, Gülseren O, Yildirim T, Ciraci S (2003) Phys Rev B 67:201401
2094, Page 8 of 8 40. Durgun E, Dag S, Ciraci S, Gülseren O (2004) J Phys Chem B 108:575 41. Kokalj A (2003) Comput Mater Sci 28(2):155 42. Everett D (2001) Manual on definitions, terminology and symbols in colloid and surface chemistry (internet version). Division of Physical Chemistry, IUPAC, Zürich 43. Cordero B, Gomez V, Platero-Prats AE, Reves M, Echeverria J, Cremades E, Barragan F, Alvarez S (2008) Dalton Trans 21:2832 44. Pyykkö P, Atsumi M (2009) Chem Eur J 15(1):186 45. Pyykkö P, Atsumi M (2009) Chem Eur J 15(46):12770 46. Xie Y, Huo YP, Zhang JM (2012) Appl Surf Sci 258:6391 47. Cuevas J, Levy Yeyati A, Martín-Rodero A (1998) Phys Rev Lett 80(5):1066
J Mol Model (2014) 20:2094 48. Cuevas J, Levy Yeyati A, Martín-Rodero A, Rubio Bollinger G, Untiedt C, Agraït N (1998) Phys Rev Lett 81(14):2990 49. Bader RFW (1994) Atoms in molecules: a quantum theory. Clarendon, Oxford 50. Henkelman G, Arnaldsson A, Jónsson H (2006) Comput Mater Sci 36:354 51. Sanville E, Kenny S, Smith R, Henkelman G (2007) J Comp Chem 28(5):899 52. Tang W, Sanville E, Henkelman G (2009) J Phys Condens Matter 21: 084204 53. IUPAC (1997) Compendium of chemical terminology. Gold book, 2nd edn. Blackwell, Oxford. doi:10.1351/goldbook, V. 2.3.2
