A density functional theory computational study of adsorption of Di-Meta-Cyano Azobenzene molecules on Si (111) surfaces

A density functional theory computational study of adsorption of Di-Meta-Cyano Azobenzene molecules on Si (111) surfaces

Applied Surface Science 422 (2017) 557–565 Contents lists available at ScienceDirect Applied Surface Science journal homepage: www.elsevier.com/loca...

2MB Sizes 0 Downloads 54 Views

Applied Surface Science 422 (2017) 557–565

Contents lists available at ScienceDirect

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

Full Length Article

A density functional theory computational study of adsorption of Di-Meta-Cyano Azobenzene molecules on Si (111) surfaces Benyamin Motevalli a , Neda Taherifar a , Bisheng Wu b,∗ , Wenxin Tang c,∗ , Jefferson Zhe Liu a,∗ a

Department of Mechanical and Aerospace Engineering, Monash University, Clayton, Victoria 3800, Australia CSIRO Energy, 71 Normanby Road, Clayton, VIC 3169, Australia c College of Materials Science and Engineering, Chongqing University, Chongqing 400044, PR China b

a r t i c l e

i n f o

Article history: Received 29 January 2017 Received in revised form 18 May 2017 Accepted 28 May 2017 Available online 30 May 2017 Keywords: Di-meta-cyano azobenzene molecule Si (111) surfaces Molecular switch Smart surfaces

a b s t r a c t The adsorption of di-meta-cyano azobenzene (DMC) cis and trans isomers on non-passivated and passivated Si (111) (7 × 7) surfaces is studied using density functional theory (DFT) calculations. Our results reveal that on the non-passivated surface the 12 Si adatoms are accessible to form chemical bonds with DMC molecules. Interestingly, the trans isomer forms two chemical bonds near the corner hole atom in Si (111) (7 × 7) surface, which is not observed in the widely studied metallic surfaces. The DMC isomers show significant structural distortion in the chemisorption case. The strong chemical bonds (and high bonding energy) could be detrimental to conformation switching between these two isomers under external stimuli. The physisorption case is also examined. Monte Carlo (MC) simulations with empirical force fields were employed to search about 106 different adsorption positions and DMC molecule orientations to identify the stable adsorption sites (up to six). The DFT-PBE and DFT-D2 calculations were then carried out to obtain the relaxed atomistic structures and accurate adsorption energy. We find that it is imperative to take van der Waals (vdW) interaction into account in DFT calculations. Our results show that the adsorption sites generally are encompassed by either the Si adatoms or the passivated H atoms, which could enhance the long-range dispersion interaction between DMC molecules and Si surfaces. The molecular structures of both isomers remain unchanged compared with gas phase. The obtained adsorption energy results Eads are moderate (0.2–0.8 eV). At some adsorption sites on the passivated surface, both isomers have similar moderate Eads (0.4–0.6 eV), implying promises of molecular switching that should be examined in experiments. © 2017 Published by Elsevier B.V.

1. Introduction Smart surfaces that can change their surface properties under external stimuli have attracted enormous attention in last decade [1–11]. Controlling the surface properties such as wettability, optical properties, thermal stability, and chemical reactivity lead to a wide range of applications, such as micro and nanofluidic devices, self-cleaning surfaces, anti-fog surfaces, molecular motors, data storage devices, medical devices, drug delivery systems, and microelectronic applications. One approach toward this goal is adsorption of molecular switches on a surface [1–5,12–14]. These switchable molecules can change their geometrical structure, spin,

∗ Corresponding authors. E-mail addresses: [email protected] (B. Wu), [email protected] (W. Tang), [email protected] (J.Z. Liu). http://dx.doi.org/10.1016/j.apsusc.2017.05.240 0169-4332/© 2017 Published by Elsevier B.V.

color, conjugation, dipole moment or other properties under external stimuli like light, electric field, electron injection, and heat [7,10,11,15,16]. Azobenzene is a family of molecules that are extensively studied as molecular switches due to their capability in reversible switching upon optical excitation. Azobenzene molecules can exist in two forms: the planar trans isomer and the cis isomer with threedimensional bent conformation. In the gas and solution phase, the trans isomer can be transformed into cis form by UV light, meanwhile the reversible switching can be induced by visible light or thermal annealing [17,18]. Recently, extensive studies have been conducted to examine the switching process and mechanism of azobenzene derivatives on different metallic materials [19–25]. Reversible switching of DMC molecules on Bi (111) surface induced by resonant X-ray illumination was observed in experiments [23,24]. Unlike the gas and solution phases, switchable molecules adsorbed on a surface are subject to restrictions aris-

558

B. Motevalli et al. / Applied Surface Science 422 (2017) 557–565

ing from the interactions between the molecules and the surface [1,26,27]. The interaction between the azobenzene derivative and substrate plays a pivotal role on the switching performance. At present, the molecular switches are either attached to the substrates via ligands [28,29] or directly deposited on substrates [19–25]. The latter option is appealing since no complex chemical process is required. It is fundamentally important to explore how the structure and the conformational dynamics of molecular switches, such as azobenzene-based compounds, are modified when directly adsorbed on solid surfaces. The previous studies of adsorbed states of azobenzene derivatives on different metallic materials, such as Au, Ag, and Cu, have identified two main physical or chemical interactions[19,26,27], (i) the tendency of the nitrogen in the azo bridge ( N N ) to form chemical bond with the surface and (ii) the long range intermolecular interaction between the phenyl ring moieties with the surface. The formation of chemical bonds shortens the distance of the molecule from the surface, often causing the Pauli repulsion force to the phenyl moiety. Hence, the cis isomer usually has a better chance to form stronger chemical bonds, due to its bent conformation. A chemical bonding results in strong adsorption energies, distortion of atomic structures, and changes of the electronic properties, which often cause loss of the switching capability. If the isomers are physisorbed, owing to its planar conformation, the trans isomer forms a stronger intermolecular van der Waals (vdW) interaction. A strong vdW interaction could cause adverse effect on the switching process [20,23,30]. However, if the interaction of the molecule with the surface is too weak, the adsorption may not be stable. Hence, it is reasonable to expect that a surface that can facilitate switching process should be the one with a balanced interaction with the molecule switch. Silicon is the most widely used substrate material in semiconductor industry and in semiconductor physics research field. There are many well-developed technologies available to manipulate and control the Si surfaces. It is promising and valuable to develop Si based smart surfaces. But there are no studies to understand how the surface adsorption modifies the conformation of DMC isomers and energetic properties of adsorption complexes. In this paper, the adsorption properties of DMC on Si (111) (7 × 7) surface and hydrogen passivated Si (111) surfaces are investigated using DFT calculations, which should provide valuable clues for experimental development of Si based smart surfaces. For the physisorption case, Grimme’s scheme (DFT-D2) was employed to account for the long-range dispersion interaction, which has an essential effect on calculated adsorption energy results [27]. To understand how the molecule interacts with the substrate, electron charge analysis was performed as well.

2. Adsorption sites on Si (111) surfaces Different from the mostly studied (111) surfaces of Au, Ag, and Cu, Si (111) (7 × 7) surface has a complex structure, which could lead to many local meta-stable adsorption sites. Locating the adsorption sites on such a complex surface through DFT calculations is a challenging task. Thus for the case of chemisorption, we examine the most plausible adsorption sites (details later). As for the physisorption case, we used MC simulations to explore the surfaces and identify possible adsorption sites. Note that the adsorption structures and energetics often depend on coverage [31,32]. We only consider the low coverage case, influence of coverage density on adsorption stability can be neglected [33]. Moreover, influence of the adsorbed molecules on Si surface structure should be minor [34].

2.1. Non-passivated surface Fig. 1a demonstrates the atomistic structure of Si (111) (7 × 7) surface. The lattice constants of Si (111) (7 × 7) within the surface plane are a = b = 27.06 Å. The four top atomic layers undergo structural reconstruction [35]. As a result, there are 19 atoms with dangling bonds and different reactivities [36]. These unsaturated atoms are classified into three groups: 12 adatoms (top layer), 6 rest atoms (2nd layer), and 1 corner hole atom (bottom layer). It is reasonable to expect that the Si atoms with dangling bonds would have a high reactivity and thus a high tendency to form chemical bonds with DMC molecule. Among them, the 12 adatoms are the most accessible unsaturated atoms since they are arranged in the top layers. Indeed, our DFT calculations confirmed that other unsaturated atoms were not accessible for DMC molecules to form a chemical bond. Hence, only the adatoms are recognized as plausible chemisorption sites on Si (111) (7 × 7) surface. These adatoms have been numbered in Fig. 1b. Note that due to the structural symmetry, some adatoms such as 2&3, 4&6, 7&9, and 10&11 are similar. In following, we will use adatom IDs to label chemisorption sites. To locate the stable physisorption regions, we scanned the Si (111) (7 × 7) surface by using Monte Carlo (MC) simulations. We expect the phenyl ring − surface interaction would be the driving adsorption mechanism in these sites. The Lennard-Jones force fields with parameters from References [37,38] were used to describe the van der Waals interactions between the DMC molecule and Si surface. We explored a large number of surface locations and molecular orientations for both cis and trans isomers. Ten runs (with different initial guesses) were carried out. In each run, following the Markov chain model, 1 × 106 random configurations were examined and the one with the lowest energy was identified. Then, the obtained structures were further minimized through DFT calculations. Fig. 1c shows the identified adsorption sites. All MC simulations yielded only one adsorption site for the cis isomer (Site I), while five different sites were obtained for the trans isomer. Site I is located on the top of the corner hole atom and surrounded by 6 adatoms (corner adatoms). Site II is surrounded by 6 adatoms (2 corner adatoms and 4 center adatoms) and encompasses one rest atom. The geometry of site III is similar to that of site II except that it is placed in the faulted half of the Si (111) (7 × 7) surface. Sites IV and VI are similar to sites III and II, respectively. They are encompassed by 4 center adatoms. Site V is located at the boundary between the two halves and is surrounded by 6 adatoms (4 corner adatoms and 2 center adatoms). Analyzing the Si (111) (7 × 7) surface structure helps us rationalize these adsorption sites. The cis isomer prefers site I because the concave hole shape of site I should well accommodate its bent molecular conformation. The N atom of cis isomer sits close to the corner hole Si atom, meanwhile the phenyl rings are near the adatoms. In Fig. 1c, all stable sites for trans isomer are flat regions confined by the adatoms. Such configurations allow DMC trans isomer to sufficiently interact with the adatoms, meanwhile the flat phenyl rings have a smaller distance from the surface. 2.2. Passivated surface The reactive dangling bonds of Si (111) (7 × 7) surface could be potentially detrimental to the switching process of the adsorbed DMC. Its reactivity can be significantly reduced via passivation. Note that at finite temperature, the Si (111) surface is often passivated [39–41]. It is known that the passivation of Si (111) (7 × 7) by hydrogen atoms depends on temperature and hydrogen concentration [40]. For a low coverage case, hydrogen atoms react with dangling bonds without changing the reconstructed surface structure, whereas a

B. Motevalli et al. / Applied Surface Science 422 (2017) 557–565

559

Fig. 1. (a) Side view of the Si (111) (7 × 7) surface. The orange atoms are the adatoms at the top layer, the red ones are the rest atoms located at the second layer, and the green one is the corner hole atom that is inside the hole. All these atoms have dangling bonds. (b) Top view of the atomistic structure of Si (111) (7 × 7) surface. The 12 adatoms are numbered. They are recognized as the plausible chemisorption sites. The unit cell is enclosed by solid lines. The dashed-line divides the unit cell in two halves, in which the left and right regions are known as unfaulted and faulted halves, respectively [36]. (c) The stable physisorption sites identified through MC simulation. All sites are enclosed by multiple adatoms. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 1 Three Si (111) surfaces with hydrogen passivation. Surface Structure

Passivation

Si–H19

Si (111) (7 × 7)

Si–H43

Reconstructed Si (111) (7 × 7) with the top adatom layer removed Si (111) 1 × 1

19 dangling bonds saturated Adatoms removed and 43 dangling bond saturated

Si–H49

49 dangling bonds saturated

high level of hydrogen atom coverage will change the surface structure by removing the adatoms or a part of them [42]. In this work, three different cases are studied. In the first case, all unsaturated atoms are passivated. In the second one, all adatoms in the top layer are removed and then all dangling bonds are saturated by 43 hydrogen atoms [43]. The third case is the fully hydrogen passivated Si (111) (1 × 1) surface [43]. To make this last case comparable to others, a supercell with a similar size as the Si (111) (7 × 7) was adopted. Table 1 provides a brief summary of these three cases, denoted as Si-H19 , Si-H43 , and Si-H49 , respectively. After hydrogen passivation, the chemisorption of DMC on these surfaces is unlikely. Hence, we employed the MC simulations to search the adsorption sites with the strongest intermolecular vdW interactions. The surface structure of Si-H19 is similar to the unsaturated Si (111) (7 × 7) surface. In this case, our MC simulations

identified sites I and V for cis and trans isomers, respectively. After passivation, the hydrogen atoms on the rest atoms force the adsorbate molecule to stand in a farther distance from the surface. As sites II, III, IV, and VI (in Fig. 1c) contain one rest atom (marked by red in Fig. 1c), they are not stable adsorption sites for trans isomer anymore. In the case of Si-H43 , the stable adsorption site for cis isomer is similar to site I of the bare Si (111) (7 × 7) surface. This site also has a concave shape that can best accommodate the bent cis isomer (Fig. 2a). As for the trans isomer, the MC simulations indicated that the strongest interactions occur in the two marked triangles in Fig. 2a. One lays in the unfaulted region and the other is in the faulted region. This could be attributed to the high density of hydrogen atoms in these triangles. In each triangle, three symmetry equivalent adsorption sites are identified (Fig. 2b). As seen in Fig. 2b, each ring of the trans isomer is confined by six surrounding hydrogen atoms, meanwhile it interacts with hydrogen atom in the center. The adsorption sites in the faulted and unfaulted regions are denoted as VII and VIII, respectively. In the case of Si-H49 , the surface is highly symmetric and the hydrogen distribution is similar to the triangles in Fig. 2a. Hence, only one distinct adsorption site is located (denoted as site IX). 3. Computational methodology The Vienna Ab initio Simulation Package (VASP 5.3.3) was used for the DFT calculations with the Perdew, Burke, and Ernzerhof

560

B. Motevalli et al. / Applied Surface Science 422 (2017) 557–565

Fig. 2. (a) Top view of the surface structure of Si-H43 . The hole, marked by violate, is the adsorption site for cis isomer. The two marked triangles have the highest density of hydrogen atoms on the surface. The solid lines show a typical adsorption region of trans isomer in these triangles. (b) There are three possible physisorption sites in the triangles. They are symmetrically equivalent.

(PBE) exchange-correlation (xc) functional (dented as DFT-PBE) and the projector augmented wave (PAW) method [44–46]. The cutoff energy of plane wave basis-set was 400 eV and the Monkhost-Pack k-point mesh was used. For the Si (111) (7 × 7) surface slab supercell, a k-point mesh of 2×2×1 was employed. A similar k-point mesh density was applied for other supercells. These settings were tested to converge the total energy values within 1 meV/atom. Previous study revealed the crucial role of vdW interaction for the molecular adsorption on solid surfaces [26]. We, therefore, also carried out DFT calculations with dispersion-corrected xc functional following Grimme’s scheme (DFT-D2).[47] Six atomic layers were included in the surface slab supercells. The four top layers are the reconstructed surface of Si (111) (7 × 7), while two more layers from the Si bulk were added at the bottom of the surface slab. These two layers were held fixed in their relaxed bulk positions and the dangling bonds were passivated by hydrogen atoms. As mentioned above, only the low coverage adsorption is considered in the present study. The unit-cell of Si (111) (7 × 7) is sufficiently large (a = b = 27.06 Å) to ensure the isolation of the adsorbates. Further, a vacuum layer with a thickness of 15 Å was used to isolate the surface slabs in c-direction. DFT calculations were performed to relax the adsorbate-surface complexes and calculate the adsorption energies. In the chemisorption case (Fig. 1b), the DMC isomers were initially located above the adtaoms at a distance of ∼ 3 Å. For the physisorption cases, the minimum energy configurations obtained from MC simulations were used as initial structures of our DFT calculations. After relaxation, the adsorption energies were calculated as Eads = Etot (DMC+Si) − Etot (DMC) − Etot (Si)

(1)

where Etot (DMC+Si), Etot (DMC), and Etot (Si) are the total energy of the adsorption complex, an isolated DMC molecule, and the Si surface, respectively. 4. Results and discussions Fig. 3 shows structures of the trans and cis isomers of DMC molecule in gas phase. The geometry of DMC molecule can be characterized by two angular parameters: the rotational angle ˇ of phenyl ring around N C bond and the dihedral angle ω that depicts the rotation of phenyl rings around nitrogen bridge ( N N ). In the gas phase, the total energy value of trans isomer is about 0.5 eV lower than that of cis isomer, indicating that the trans configuration

is more stable in the gas phase [21]. Another important geometric parameter is the N N bond length, about 1.27 Å and 1.25 Å in the gas phase for trans and cis isomers, respectively. The changes of these parameters and the distances between DMC and surface could provide some clues to understand adsorption state. Slight geometric distortion and a relatively large distance (about 3 Å) could suggest physisorption. It is likely to be chemisorption if the distortions are significant and DMC stands in a short distance from the surface (about 2 Å).

4.1. Chemisorption of DMC molecule on non-passivated Si (111) (7 × 7) surface Both isomers were initially placed above the 12 adatoms at a distance of ∼ 3 Å. After relaxation in our DFT calculations, the calculated adsorption energies and the corresponding relaxed geometric parameters are summarized in Table 2. Significant structural distortion and relatively higher adsorption energy results are observed. The N-Si distance (the nearest distance between the nitrogen atom and the adatoms) is reduced to around 1.8 Å. The isomers undergo significant distortions compared with their gas phase. In Table 2, the average ( N N ) bond elongation for both isomers is quite similar (8.7%). The N Si bond in cis isomer is slightly shorter (about 3%) than in trans isomer, which could be attributed to the bent conformation of cis isomer. Overall, it is observed that a shorter N-Si distance results in a higher elongation of the N N bond and a stronger N-Si bond could weaken the N N bond, leading to a longer bond length. Regarding the torsion angle, it increases from 0◦ (gas phase) to ∼20◦ −30◦ for trans isomers, whereas this change is from 12◦ (gas phase) to around 45◦ for the cis isomer. Fig. 4 demonstrates the relaxed complexes at adatom 1. The structural distortion of the trans isomer is more evident. One of the rings (the left one in Fig. 4a) has tilted upward to stand in a farther distance from the surface. The other ring is located quite close to the surface with the nearest carbon-adatom distance of 2.07 Å. This is due to formation of two chemical bonds, N-Si and C5-Si (more evidences later). In the case of cis isomer, the distortion mainly comes from the C N N C bonds. The nitrogen atom bonded to the Si adatom is displaced, causing distortions on the C N N C bonds. A significant change in the ω torsion angle is observed. The cis isomer preserves its bent structure with a slightly large bent angle compared with its gas phase (Fig. 4b).

B. Motevalli et al. / Applied Surface Science 422 (2017) 557–565

561

Fig. 3. The conformation of trans and cis isomers of DMC in gas phase and their relative total energy difference obtained from our DFT-D2 calculations. This energy difference is 0.59 eV by using DFT-PBE calculations, which is consistent with the result in Ref. [19].

Table 2 The chemisorption energy Eads and geometries of relaxed adsorption complex at different adsorption sites (labeled by adatom IDs) on Si (111) (7 × 7) surface. Adatom

Eads (eV)

1 2&3 4&6 5 7&9 8 10&11 12 Gas phase

Trans −1.23 −0.67 −1.23 −0.83 −1.24 −0.88 −0.71 −1.24 –

Trans cis −1.11 −1.16 −1.06 −1.17 −1.22 −1.22 −1.23 −1.17 –

N-Si (Å) 1.84 1.88 1.84 1.87 1.84 1.87 1.88 1.83 –

cis N N (Å) 1.398 1.354 1.397 1.368 1.385 1.369 1.357 1.398 1.267

ω (deg) 31.8 18.7 30.9 19.3 30.5 19.8 18.7 31.0 0

N-Si (Å) 1.79 1.82 1.81 1.80 1.80 1.80 1.81 1.79 –

N N (Å) 1.368 1.354 1.365 1.359 1.360 1.361 1.358 1.372 1.253

ω (deg) 49.5 40.7 47.4 44.8 44.0 45.3 42.3 50.5 11.9

Fig. 4. The relaxed structures of a DMC molecule chemisorbed at adatom 1 on Si (111) (7 × 7) surface: (a) trans isomer and (b) cis isomer. Two chemical bonds are formed in (a).

The adsorption energy values of cis isomer at all adatoms are higher than 1 eV. The adsorption in the unfaulted half is slightly stronger (about 6% higher) than that in the faulted half, suggesting that the unfaulted half is more reactive. In the case of trans isomer, the adsorption energy shows a considerable variation. Interestingly, we found that at adatoms surrounding the hole atoms, the adsorption energy is high (about 1.24 eV) and comparable to cis isomer. The strong chemical binding of trans isomer is different from other studies on metallic surfaces, in which the cis isomer exhibits chemisorption [19,26]. For instance, Henningsen et al., [19] studied adsorption of DMC molecule on Cu (100) surface. They obtained the adsorption energies of trans and cis isomers as 0.29 eV and

1.03 eV, respectively. The N Cu interatomic distance for cis isomer was about 2 Å and its structure was significantly distorted. They attributed the relatively low adsorption energy to the repulsion between the phenyl rings and surface. A similar observation was reported in Ref. [48], in which they examined adsorption of Azobenzene (AB) molecule on Cu (111). The variation in the adsorption energy could be explained by the atomistic structure of the Si (111) (7 × 7) surface and analysing the chemical or physical interactions of the complexes. Once chemisorption occurs, as described by Ref. [27], the adsorption energy could be characterized by three main components: (i) the adsorption energy due to the formation of covalent-type bonds

562

B. Motevalli et al. / Applied Surface Science 422 (2017) 557–565

between the molecule and surface, (ii) the strain energy arising from the molecular distortion of the adsorbate, and (iii) the Pauli repulsion of the phenyl rings and the surface. While the first component is in favour of strong binding, the other two are unfavourable. Considering the flat structure of trans isomer, the third effect is significant, which prevents the molecule from getting close to the surface. As a result of the repulsion, the adsorbate undergoes distortion where the rings are usually tilted upward. For this reason, as seen in Table 2, the adsorption energies are often lower than those of cis isomers. There is an exception for trans isomer at the adatoms surrounding the hole. Our electronic charge analysis (see Section D) revealed that two chemical bonds are formed at these adatoms. The second bond is associated with the C5 carbon atom and one of the neighbouring adatoms on the hole. Clearly, the complex Si (111) surface provides an opportunity for the rings to get closer to the adatoms (and thus form a chemical bond) meanwhile experiences less repulsion from other surface atoms. The bent structure of cis isomer leads to minor Pauli repulsion. Therefore, only slight variation is observed for cis at different adatoms. We also examined the possibility of chemisorption near the hole and rest atoms (with dangling bonds). The hole atom is located deep inside the surface and is completely out of reach for DMC isomers. The rest atoms lay in the 2nd top layer and could be potentially more accessible. We placed the isomers on top of the rest atoms and employed DFT calculation to relax the complex. Both cis and trans isomers moved upward and the N-Si distance increased from 3 Å to about 4 Å. The calculated adsorption energies were very low (e.g., −0.02 eV for cis and −0.15 eV for trans) and their structures remained intact. These evidences indicate that the rest atoms are not accessible for chemisorption of DMC molecules. In a short summary, the DMC molecule can be chemically adsorbed to adatoms on Si (111) (7 × 7) surface. The trans isomers are significantly distorted and notable distortion is observed for cis isomers. The Eads value is on the order of 1 eV. The cis isomer has quite similar Eads at different adatoms, whereas the trans isomer shows a notable variation at different adatom sites. This could be attributed to its flat molecular structure. Interestingly, at some adatom sites, trans isomer apparently forms an additional chemical bond with a second Si adatom. Based on previous studies, such a strong chemical interaction could be detrimental to switching between trans and cis isomers. 4.2. Physisorption of DMC molecules on non-passivated Si (111) (7 × 7) surface The comprehensive MC simulations identified six different stable physisorption sites (Fig. 1c), among which sites I and II have the strongest intermolecular interaction with cis and trans isomers, respectively. DFT calculations were then employed to relax the MC configurations and obtain the adsorption energies. Our calculations concluded that the presence of surface had a negligible effect on the structural geometry of the adsorbed isomers. The trans isomers largely keep its planar structure. The cis isomers have the dihedral angle ω close to 12◦ , which is similar to the gas phase, and the nitrogen–nitrogen bond lengths ( N N–) do not show notable changes either. In addition, the isomers stand at a distance around 3.2 Å above the surface, which is certainly too long for a chemical bond. These evidences confirm the physisorption nature of DMC molecules at these six sites [21,26]. The calculated Eads results are summarized in Fig. 5a. The negative values indicate that the adsorptions are exothermic; hence, a larger negative value represents a higher affinity to the surface. Our calculations reveal that trans isomer exhibits stronger physisorption at all sites of Si (111) (7 × 7) surface. This should be attributed to the planar molecular structure of trans isomer, which could allow stronger intermolecular interactions with the largely flat region of

Fig. 5. The adsorption energy at six identified sites (MC simulations) on nonpassivated Si (111) (7 × 7), employing (a) PBE and (b) DFT-D2 xc functional, respectively.

the surface. As expected, the weakest site for trans isomer is site I with a concave shape. Site II shows the highest affinity for the trans isomer, which is consistent with our MC simulation. However, for cis isomer, site V has the strongest physisorption, which is different from our MC simulations (i.e., site I). Despite that Eads is negative at all sites, the magnitude is small (in particular, in the case of cis isomer). It is well-known that the standard PBE calculations underestimate the long-range dispersive (van der Waals) attraction, particularly in the gas-adsorbent interaction [27,49]. McNellis et al. thoroughly discussed the importance of dispersive vdW interactions in DFT calculations for the case of AB adsorption on Au, Ag, and Cu (111) surfaces [27]. After including dispersive interactions, they observed a significant increase in adsorption energies. For example, the adsorption energy of ABtrans on Ag (111) increased from 0.11 to 2.2 eV. This number changed from 0.42 to 1.58 eV in the case of AB-cis [27]. As expected, dispersion interaction has a more profound effect on trans isomer. They performed similar calculations for adsorption of AB on Cu (111) surface and observed a similar increase of adsorption energies. In light of these, we also performed DFT-D2 calculations and observed significantly increased magnitudes of Eads (Fig. 5b). In experiments, adsorption energy of ∼ 0.6 eV was obtained for benzene on Ag (111) [50], Au (111) [51], and Cu (111) [52] surfaces. Our DFT-D2 calculations yield Eads value for trans isomer on the order of 0.6–0.8 eV. In light of the similarity between the benzene molecules and phenyl moieties in DMC, we can conclude the indispensible role of the dispersion interaction in DFT calculations for our systems. Regarding the trans isomer, the variation trend of Eads at different sites is quite similar to DFT-PBE. Site II is recognized with the highest affinity with the Eads of around 0.82 eV (compared with 0.17 eV from DFT-PBE calculations). For the cis isomer, the trend of Eads on different sites is completely changed. Interestingly, our results show that site I has the strongest affinity, consistent with our MC simulations. It is reasonable since the concave surface structure of site I should provide a better chance for cis isomer to interact with the surface atoms. The adsorption energy at this site is almost two times higher than those of other sites. Note that the adsorption energies of cis and trans isomers at site I are comparable, 0.57 eV and 0.59 eV, respectively. We also found that although the inclusion of dispersion correction in DFT calculations affects the Eads

B. Motevalli et al. / Applied Surface Science 422 (2017) 557–565

563

Table 3 Total charge transferred from molecule to the surface.

Fig. 6. The adsorption energies calculated based on DFT-D2 for trans and cis isomers on different hydrogen passivated Si (111) surfaces.

considerably, it has a slight effect on the atomistic structures of the gas-adsorbent complex. In a short summary, multiple adsorption sites were identified for the physisorption case. The DMC isomers remain similar atomistic structure as those in gas phases. Our DFT-D2 calculations yield similar trend of Eads with the MC simulations. We conclude that the standard PBE calculation is inadequate for this system. 4.3. Physisorption of DMC molecule on passivated Si (111) surfaces It is highly unlikely for chemisorption since all surface dangling bonds are saturated. We employed MC simulations to identify the adsorption sites on the three passivated Si (111) surfaces (Table 1). Then DFT-D2 calculations were carried out to relax the configurations obtained from MC simulations. For Si-H19 surface, sites I and V have the strongest affinities for cis and trans isomers, respectively, which is consistent with our MC predictions. Similar to the non-passivated surface, this could be attributed to the surface structure of Si (111) (7 × 7) at these sites. The cis isomer at site I has a slight increase of Eads by 0.05 eV after passivation. However, the passivation has weakened the interaction between the trans isomer and surface. The reduction of Eads is about 0.36 eV and 0.16 eV at site I and V, respectively. Note that reduction in adsorption energies was expected since after passivation the adsorbate is forced to move farther away from the surface. trans = −0.67 eV) is Nevertheless, the adsorption energy at site V (Eads in reasonable high (i.e., much higher than thermal energy at room temperature kB T = 0.026 eV) and thus it is expected trans isomer would be thermodynamically stable at this site under ambient condition. Similarly, it is expected that site I would be a stable site for trans = −0.63) under ambient condition. adsorption of cis isomers (Eads In Fig. 2, there are only three distinct adsorption sites on SiH43 surface. Site I has a hole structure similar to non-passivated surfaces. The cis isomer has Eads value of 0.47 eV, which is only moderately reduced in comparison with the non-passivated case. At the other two sites (VII and VIII), the adsorption energy is about 0.2 eV. Note that the arrangement of the hydrogen atoms on the surface is identical at these two sites, the only difference is they lay in two different halves (see Section 2). It is expected that the cis isomer would be thermodynamically stable at site I at room temperature. As expected, the weakest affinity of trans is at site I (about 0.4 eV), while sites VII and VIII have stronger interactions with Eads values of about 0.55 eV and 0.60 eV, respectively. The Eads is slightly higher in site VIII since it is located in the unfaulted half. This could be attributed to a higher density of atoms in the top layers of this half. The trans isomer might be stable at all three sites (in Fig. 6b) at room temperature. The Si-H49 surface case is simple. There is only one adsorption site IX. The hydrogen arrangement at this site is similar to that at sites VII and VIII in the Si-H43 surface case. The adsorption energy

Case

Charge (e− )

cis@adatom 1 trans@adatom 1 cis@site I trans@site II

1.273 2.180 0.078 0.125

for cis or trans is around 0.2 eV or 0.52 eV, comparable to site VII or VIII in Si-H43 surface. The relatively low adsorption energy for cis might have thermodynamic stability issue. In a short summary, no chemisorption is observed on the passivated Si surfaces. The distance of the physically adsorbed DMC molecule from the passivation hydrogen atoms is around 2.8 Å. The DMC molecules largely remain their molecular structures as in the gas phase. The Eads value is generally reduced in comparison with the physisorption case in non-passivated Si (111) surface. In some cases, the reduction is quite small. The moderate Eads value, particularly at site I in Si-H43 surface, suggests potential molecular switching. Nevertheless, experiments should be carried out to examine this postulation. 4.4. Electronic charge analysis This section provides the electronic charge transfer analysis to gain insights into the interaction between the molecules and surface. Four typical chemisorption or physisorption cases are presented. The first one is cis chemisorbed at adatom 1 (Fig. 7a). The second case is for trans isomer chemically adsorbed in the vicinity of adatom 1 (Fig. 7b). Note that we selected this adatom to demonstrate that multiple chemical bonds could form between trans isomer and Si surface. The third and fourth cases are for cis and trans isomers physisorbed at site I and II, respectively (Fig. 7c and d). They are selected because they have the largest Eads values. Fig. 7 shows the difference-electron density results for the four cases. Significant charge redistribution between cis isomer and Si surface can be observed in Fig. 7a. Together with the short interatomic distance 1.79 Å, we can conclude the formation of a chemical bond between the nitrogen atom in the N N bridge and adatom 1. This observation is analogous to those DMC@metal-surface cases. The chemical bond also affect the electronic structure of Si atoms that are located below the adatom 1, where considerable charge redistribution can be observed as well. Interestingly, for trans isomer (Fig. 7b), the difference-electron density reveals the formation of two chemical bonds. One of the bonds is between the nitrogen atom and adatom 1, while the other is between the C5 carbon atom in the phenyl ring and Si adatom 9 (interatomic distance of 2.07 Å in Fig. 4a). Indeed, the adatoms around the hole are distributed such that the C5 carbon atom is located close enough to the neighbouring adatom 1 to form a chemical bond. As shown in Fig. 4, only the ring on the left is tilted upward owing to the Pauli repulsion. The ring on the right is in a close distance (∼ 2.07 Å) from the surface owing to the formation of the chemical bond between C5 and adatom 9. This appears to be a feature not reported before. In Fig. 7c and d, the charge difference is negligible when comparing with the chemisorption (Fig. 7a and b). This further confirms the physisorption nature at these sites. The molecules are mainly adsorbed via the long-range dispersion attraction. To quantify the amount of charge transferred to the Si surfaces, the Bader analysis is employed. [53] The results are summarized in Table 3. The transferred charge of the chemisorbed trans isomer is about twice of that of the chemisorbed cis isomer. This could be attributed to formation of two chemical bonds between trans isomer and two Si adataoms. As expected, the charge transfer in the physisorption cases is negligible.

564

B. Motevalli et al. / Applied Surface Science 422 (2017) 557–565

Fig. 7. The difference-electron density for chemisorption case: (a) cis isomer at adatom 1, (b) trans isomer in the vicinity of adatom 1; and for physisorption case: (c) cis@site I, and (d) trans@site II. The blue iso-surface represents electron accumulation and the yellow one shows the hole density. The iso-surfaces represent the identical density value of 0.001 e− /bohr3 . (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

5. Conclusion In this paper, DFT calculations are carried out to study the adsorption of DMC molecules on Si surfaces. Both non-passivated (111) (7×7) and passivated (111) surfaces were examined. On the non-passivated surface, the Si-adatoms located at the top surface layer were accessible to form chemical bonds with DMC molecules. Similar to the metal surfaces, a chemical bond is formed between one of the nitrogens in the azo bridge and the surface atoms. Interestingly, we found that the trans isomer forms a second chemical bond between a carbon atom in the phenyl ring (C5) and the adatom 9 when it is adsorbed in site I. The formation of chemical bonds changes the molecular conformation of both trans and cis isomers. One phenyl ring of trans isomer is largely parallel to the surface and the other one tilts upward from the surface owing to the Pauli repulsion. For the cis isomer, the dihedral angle ω is significantly increased from 12◦ (gas phase) to above 40◦ . The physisorption case is much more complicated. MC simulations with empirical force field were employed to efficiently explore the surface for stable adsorption sites, followed by careful DFT calculations and electron charge transfer analysis. Multiple stable adsorption sites were identified (Fig. 1). We found that it is imperative to include dispersion effect in DFT calculations for our systems. The molecular conformation of both isomers largely remains intact. The Eads values are moderate (Fig. 5), which could be a positive sign for molecular switch applications. We also examined the adsorption of DMC on passivated Si (111) surfaces. Three different passivation patterns, denoted by Si-H19 , Si-H43 , and Si-H49 were investigated. The passivation reduces the

reactivity of the surface; thus, only physisorption could take place in these cases. At all physisorbed sites except site I, trans isomer has a much higher adsorption energy value than cis isomer, which could be attributed to its planar configuration that could promote the van der Waals interactions with Si surfaces. Note that site I is special for cis isomer due to its concave hole shape that can best accommodates the bent isomer conformation. The passivation generally reduces the Eads values. The moderate adsorption energy in some cases may suggest a good opportunity for molecular switching applications, which is worth of further experimental investigation.

Acknowledgements The authors thank the support from Australian Research Council and the seed grant from Engineering Faculty of Monash University. The simulations were done using high performance computing facility from National Computational Infrastructure in Australia.

References [1] K.-Y. Chen, S.J. Wezenberg, G.T. Carroll, G. London, J.C.M. Kistemaker, T.C. Pijper, B.L. Feringa, Tetrapodal Molecular Switches and Motors: Synthesis and Photochemistry, The Journal of Organic Chemistry 79 (2014) 7032–7040. [2] R. McCreery, A. Bergren, Surface functionalization in the nanoscale domain, in: M. Stepanova, S. Dew (Eds.), Nanofabrication, Springer, Vienna, 2012, pp. 163–190. [3] Y. Pei, J. Ma, Electric field induced switching behaviors of monolayer-Modified silicon Surfaces: Surface designs and molecular dynamics simulations, J. Am. Chem. Soc. 127 (2005) 6802–6813.

B. Motevalli et al. / Applied Surface Science 422 (2017) 557–565 [4] A.-M. Resmerita, E. Epure, S. Grama, C. Ibanescu, N. Hurduc, Photochromic behaviour of nano-Structurable azo-Polysiloxanes with potential application in biology, Open Chem. Biomed. Meth. J. 2 (2009) 91–98. [5] P.M. Mendes, Stimuli-responsive surfaces for bio-applications, Chem. Soc. Rev. 37 (2008) 2512–2529. [6] B.S. Gallardo, V.K. Gupta, F.D. Eagerton, L.I. Jong, V.S. Craig, R.R. Shah, N.L. Abbott, Electrochemical principles for active control of liquids on submillimeter scales, Science 283 (1999) 57–60. [7] J. Lahann, S. Mitragotri, T.-N. Tran, H. Kaido, J. Sundaram, I.S. Choi, S. Hoffer, G.A. Somorjai, R. Langer, A reversibly switching surface, Science 299 (2003) 371–374. [8] S. Mettu, M.K. Chaudhury, Motion of drops on a surface induced by thermal gradient and vibration, Langmuir 24 (2008) 10833–10837. [9] Q. Yuan, Y.-P. Zhao, Precursor film in dynamic wetting, electrowetting, and electro-Elasto-Capillarity, Phys. Rev. Lett. 104 (2010) 246101. [10] C.L. Feng, Y.J. Zhang, J. Jin, Y.L. Song, L.Y. Xie, G.R. Qu, L. Jiang, D.B. Zhu, Reversible wettability of photoresponsive fluorine-Containing azobenzene polymer in langmuir-Blodgett films, Langmuir 17 (2001) 4593–4597. [11] K. Ichimura, S.-K. Oh, M. Nakagawa, Light-Driven motion of liquids on a photoresponsive surface, Science 288 (2000) 1624–1626. [12] N.L. Abbott, C.B. Gorman, G.M. Whitesides, Active control of wetting using applied electrical potentials and self- assembled monolayers, Langmuir 11 (1995) 16–18. [13] X. Zhu, Q. Yuan, Y.-P. Zhao, Capillary wave propagation during the delamination of graphene by the precursor films in electro-elasto-capillarity, Sci. Rep. 2 (2012). [14] W. Xiong, J. Liu, Z.-L. Zhang, Q.-S. Zhen, Control of surface wettability via strain engineering, Acta Mech. Sin. 29 (2013) 543–549. [15] B.L. Feringa, Molecular Switches, 42, Wiley Online Library, 2001. [16] S. Venkataramani, U. Jana, M. Dommaschk, F. Sönnichsen, F. Tuczek, R. Herges, Magnetic bistability of molecules in homogeneous solution at room temperature, Science 331 (2011) 445–448. [17] D.L. Beveridge, H.H. Jaffé, The electronic structure and spectra of cis- and trans-Azobenzene1, J. Am. Chem. Soc. 88 (1966) 1948–1953. [18] Z. Sekkat, W. Knoll, Photoreactive Organic Thin Films, Academic Press, 2002. [19] N. Henningsen, R. Rurali, K. Franke, I. Fernández-Torrente, J. Pascual, Trans to cis isomerization of an azobenzene derivative on a Cu (100) surface, Appl. Phys. A 93 (2008) 241. [20] N. Henningsen, K.J. Franke, I.F. Torrente, G. Schulze, B. Priewisch, K. ´ T. Klamroth, P. Saalfrank, J.I. Pascual, Inducing the Rück-Braun, J. Dokic, rotation of a single phenyl ring with tunneling electrons, J. Phys. Chem. C 111 (2007) 14843–14848. ´ P. Saalfrank, On the electronic structure of [21] G. Füchsel, T. Klamroth, J. Dokic, neutral and ionic azobenzenes and their possible role as surface mounted molecular switches, J. Phys. Chem. B 110 (2006) 16337–16345. [22] C. Bronner, P. Tegeder, Photo-Induced and thermal reactions in thin films of an azobenzene derivative on Bi (111), New J. Phys. 16 (2014) 053004. ¨ [23] C. Bronner, B. Priewisch, K. Ruck-Braun, P. Tegeder, Photoisomerization of an azobenzene on the Bi (111) surface, J. Phys. Chem. C 117 (2013) 27031–27038. [24] A. Krüger, M. Bernien, C.F. Hermanns, W. Kuch, X-ray-induced reversible switching of an azobenzene derivative adsorbed on Bi(111), J. Phys. Chem. C 118 (2014) 12916–12922. [25] R.J. Maurer, K. Reuter, Bistability loss as a key feature in azobenzene (Non− ) switching on metal surfaces, Angew. Chem. Int. Ed. 51 (2012) 12009–12011. [26] E. McNellis, J. Meyer, A.D. Baghi, K. Reuter, Stabilizing a molecular switch at solid surfaces: a density functional theory study of azobenzene on Cu(111), Ag(111), and Au(111), Phys. Rev. B 80 (2009) 035414. [27] E.R. McNellis, J. Meyer, K. Reuter, Azobenzene at coinage metal surfaces: role of dispersive van der waals interactions, Phys. Rev. B 80 (2009) 205414. [28] R. Schmidt, E. McNellis, W. Freyer, D. Brete, T. Gießel, C. Gahl, K. Reuter, M. Weinelt, Azobenzene-Functionalized alkanethiols in self-Assembled monolayers on gold, Appl. Phys. A 93 (2008) 267–275. [29] H. Jacob, S. Ulrich, U. Jung, S. Lemke, T. Rusch, C. Schütt, F. Petersen, T. Strunskus, O. Magnussen, R. Herges, Monitoring the reversible photoisomerization of an azobenzene-Functionalized molecular triazatriangulene platform on Au (111) by irras, Phys. Chem. Chem. Phys. 16 (2014) 22643–22650.

565

[30] N. Henningsen, R. Rurali, C. Limbach, R. Drost, J.I. Pascual, K.J. Franke, Site-Dependent coordination bonding in self-Assembled metal-Organic networks, J. Phys. Chem Lett. 2 (2011) 55–61. [31] R.J. Maurer, W. Liu, I. Poltavsky, T. Stecher, H. Oberhofer, K. Reuter, A. Tkatchenko, Thermal and electronic fluctuations of flexible adsorbed molecules: azobenzene on Ag (111), Phys. Rev. Lett. 116 (2016) 146101. [32] G. Mercurio, R.J. Maurer, W. Liu, S. Hagen, F. Leyssner, P. Tegeder, J. Meyer, A. Tkatchenko, S. Soubatch, K. Reuter, Quantification of finite-Temperature effects on adsorption geometries of -Conjugated molecules: azobenzene/Ag (111), Phys. Rev. B 88 (2013) 035421. [33] M. Willenbockel, R.J. Maurer, C. Bronner, M. Schulze, B. Stadtmüller, S. Soubatch, P. Tegeder, K. Reuter, F.S. Tautz, Coverage-Driven dissociation of azobenzene on Cu (111): a route towards defined surface functionalization, Chem. Commun. 51 (2015) 15324–15327. [34] K. Oura, A. Zotov, V. Lifshits, A. Saranin, M. Katayama, Surface Science, Springer, 2003. [35] M. Smeu, H. Guo, W. Ji, R.A. Wolkow, Electronic properties of Si (111)-7 × 7 and related reconstructions: density functional theory calculations, Phys. Rev. B 85 (2012) 195315. [36] K. Takayanagi, Y. Tanishiro, M. Takahashi, S. Takahashi, Structural analysis of Si(111)-7 × 7 by uhv-Transmission electron diffraction and microscopy, J. Vacuum Sci. Technol. A 3 (1985) 1502–1506. [37] G.M. Morris, D.S. Goodsell, R.S. Halliday, R. Huey, W.E. Hart, R.K. Belew, A.J. Olson, Automated docking using a lamarckian genetic algorithm and an empirical binding free energy function, J. Comput. Chem. 19 (1998) 1639–1662. [38] E.R. Cruz-Chu, A. Aksimentiev, K. Schulten, Water-Silica force field for simulating nanodevices, J. Phys. Chem. B 110 (2006) 21497–21508. [39] A. Vittadini, A. Selloni, Binding sites, migration paths, and barriers for hydrogen on Si (111)-(7 × 7), Phys. Rev. Lett. 75 (1995) 4756. [40] R.-L. Lo, M.-S. Ho, S. Hwang, T.T. Tsong, Diffusion by bond hopping of hydrogen atoms on the Si (111)-7 × 7 surface, Phys. Rev. B 58 (1998) 9867. [41] R.-L. Lo, I.-S. Hwang, M.-S. Ho, T.T. Tsong, Diffusion of single hydrogen atoms on Si(111)-(7 × 7), Phys. Rev. Lett. 80 (1998) 5584–5587. [42] C. Beitia, W. Preyss, R. Del Sole, Y. Borensztein, Adsorption kinetics of H on Si (111) 7 × 7 by means of surface differential reflectivity, Phys. Rev. B 56 (1997) R4371. [43] D. Alfonso, C. Noguez, D. Drabold, S. Ulloa, First-principles studies of hydrogenated Si (111)-7 × 7, Phys. Rev. B 54 (1996) 8028. [44] Q. Sun, Q. Wang, P. Jena, Y. Kawazoe, Clustering of Ti on a C60 surface and its effect on hydrogen storage, J. Am. Chem. Soc. 127 (2005) 14582–14583. [45] G. Kresse, D. Joubert, From ultrasoft pseudopotentials to the projector augmented-wave method, Phys. Rev. B 59 (1999) 1758. [46] J.P. Perdew, K. Burke, M. Ernzerhof, Generalized gradient approximation made simple, Phys. Rev. Lett. 77 (1996) 3865. [47] S. Grimme, Semiempirical gga-type density functional constructed with a long-range dispersion correction, J. Comput. Chem. 27 (2006) 1787–1799. [48] E. McNellis, J. Meyer, A.D. Baghi, K. Reuter, Stabilizing a molecular switch at solid surfaces: a density functional theory study of azobenzene on Cu (111), Ag (111), and Au (111), Phys. Rev. B 80 (2009) 035414. [49] J. Shang, G. Li, R. Singh, P. Xiao, D. Danaci, J.Z. Liu, P.A. Webley, Adsorption of co2, N2, and ch4 in Cs-exchanged chabazite: a combination of van der waals density functional theory calculations and experiment study, J. Chem. Phys. 140 (2014) 084705. [50] X.-L. Zhou, M. Castro, J. White, Interactions of Uv photons and low energy electrons with chemisorbed benzene on Ag (111), Surf. Sci. 238 (1990) 215–225. [51] D. Syomin, J. Kim, B.E. Koel, G.B. Ellison, Identification of adsorbed phenyl (C6h5) groups on metal surfaces: electron-induced dissociation of benzene on Au (111), J. Phys. Chem. B 105 (2001) 8387–8394. [52] M. Xi, M.X. Yang, S.K. Jo, B.E. Bent, P. Stevens, Benzene adsorption on Cu (111): formation of a stable bilayer, J. Chem. Phys. 101 (1994) 9122–9131. [53] M. Yu, D.R. Trinkle, Accurate and efficient algorithm for bader charge integration, J. Chem. Phys. 134 (2011) 064111.