First-principles study of the halide-passivation effects on the electronic structures of CdSe quantum dots

Abstract

We studied the effects of chlorine passivation and iodine passivation on the electronic structures of Cd₃₃Se₃₃ quantum dots through partial chlorine replacement for surface pseudo-hydrogen atoms, taking full pseudo-hydrogen-terminated Cd₃₃Se₃₃ quantum dots as a reference. Our calculations demonstrate that the electrostatic interaction between surface Cd absorbates and the halide passivant removes the dangling-bond-derived states of surface Cd atoms. Due to the high electronegativity, the Cl passivants need to coordinate with three Cd atoms to saturate p orbitals and achieve complete saturation. The modulation of the Cl passivants to the electronic structures of quantum dots depends on the coordination number of the Cl passivants. With coordination numbers of up to three, Cl passivants contribute less to the HOMO state and the QDs are more energetically stable. As the electronegativity decreases from Cl to I, I passivants with a coordination number of 2 are energetically stable enough to passivate QDs, leading to a relatively inactive surface after passivation. The HOMO states of I-terminated QDs are composed of I 5p and Se 4p.

---

1 Introduction

Colloidal semiconductor quantum dots (QDs), because of the quantum size effect and strain relaxation, have unique physical properties such as tunable energy band-gaps, carrier separation and size-dependent optical properties.^1–3 Their tunability enables the use of QDs to optimally match the optical absorption of solar cells to the solar spectrum. In this regard, QDs are considered as one of the most promising materials for photovoltaic devices of the future. The high surface-to-volume ratio of QDs leads to the existence of a large number of surface dangling bonds which inject mid-gap states into the band gap. The open-circle voltage, V_oc, which is established by two distinct quasi-quantum energy levels, one for electrons and the other for holes, influences the performance of the solar cells. When the dangling bonds introduce trap states, the photogenerated charges fill into the mid-gap levels instead of the conductive band levels, resulting in the reduction of V_oc. Moreover, the dangling bonds serve as the centers of electron–hole recombination, shortening the life of excitons, which has a negative effect on the short-circuit current, J_sc. Consequently, the performances of nanostructured-semiconductor-based photovoltaics are limited by the surface states derived from dangling bonds. To improve solar cell efficiency, a strategy that achieves maximal passivation levels is desired to remove the effect of surface dangling bonds.

Experiments have shown that surface coordinating ligands have a dramatic effect on the band edges.^4–9 For example, the rate of intra-band relaxation can be slowed down by changing the capping organic ligands.⁴ Organic alkylamines⁹ and inorganic multi-semiconductor-shells^9,10 increase the emission quantum yields by passivating the surface of CdSe QDs. Surface passivants also influence multiple-exciton generation efficiencies.⁸ Surface passivation with appropriate ligands improves the quantum efficiency of band edges.⁹ The use of short mercaptopropionic acid for organic ligand passivation achieved the best reported efficiency of 5.6% by reducing the organic ligand interparticle spacing.⁷ Much attention has been devoted to organic ligands which are used to lower the recombination loss due to the surface dangling bonds.^4,7,11,12 However, the bulkiness of the organic ligands that employ long chains (8–18 carbons) prevent organic passivants from passivating other cations on the surface of QDs, leading to low coverage of passivants. The incomplete coverage of organic ligands leaves a high density of defects which leads to recombination loss. Monovalent inorganic ligands⁵ were recently reported to offer a better passivation considering steric effects, achieving efficiencies as high as 6.0%. Halide ions, which attach to CdSe QDs through electrostatic interactions, lead to different degrees of redispersion and stabilization in polar solvents.¹³ Moreover, a hybrid passivation scheme⁶ of short organic ligands and halide anions has improved the efficiencies of colloidal QD solar cells, achieving as high as 7%.

First principles calculations have showed that organic ligands form delocalized orbitals spreading over the QD surface and ligands.^12,14 Steric interference of neighboring organic ligands affects the binding strength between oxygen atoms in the organic ligands and cations on the surface.¹¹ Unlike the well-studied effects of organic ligands on both structure and physical properties, the effects of inorganic ligands are rarely studied.^15–17 In theoretical studies of nanostructured semiconductors, pseudo-hydrogen (PH) has been proposed to passivate each of the dangling bonds with a fractional charge of (8 − Z)/4, where Z is the valence charge of the surface atom.¹⁸ With fractional charge, the PH passivants saturate the dangling bonds and remove surface states from the electronic structure of the band edges.¹⁹ However, PH passivants are artificial and do not exist in the real world. There is no way of getting the same intrinsic electronic structure as in the PH passivation case. However, we can take PH passivation as a reference, which is an ideal strategy to remove the dangling bonds. Experiments have shown that atomic halide ions, with the goal of using the shortest imaginable ligands, passivated QDs.^5,6,13 However, the coordination effects of halide anions on the electronic structure of QDs have still not been investigated. Herein, we report results of density function theory (DFT) calculations that reveal the effects of chlorine passivation on the morphology and electronic structure of CdSe QDs. We used Cd₃₃Se₃₃ clusters, which have been experimentally identified to be one of the smallest CdSe clusters.²⁰

In this paper we investigated the different impacts of PH and halide (Cl and I) passivations on the electronic properties of passivated QDs. For that purpose, we focused on Cd₃₃Se₃₃ dots capped with chlorine or iodine atoms on topical sites and analyzed the changes in the corresponding electronic structure. We show that the electronic interaction between surface Cd absorbates and halide passivants can remove the surface states from dangling bonds of surface Cd atoms. However, the incompletely saturated Cl passivants affect the states of the highest occupied molecular orbitals (HOMO). Therefore, the effect of chlorine passivation depends on the coordination number of the Cl passivants. With a coordination number of 3, the Cl passivants remove surface states and have a negligible effect on the HOMO states. In iodine passivation, the 2-coordinated I passivants are energetically stable and have energy levels close to those of Cd–Se in QDs, which greatly affects the HOMO states.

2 Methods

In our DFT calculations, we carried out electronic structure and total energy calculations using the Troullier–Martins pseudopotentials prescription and the Perdew–Burke–Ernzerh of generalized-gradient approximation (GGA) exchange-correlation potential in the Siesta package. Double-ζ plus polarization orbitals (DZP) are used as a basis set. The initial quantum dot coordinates are obtained by cutting out from the bulk wurtzite lattice (a = 4.419 Å, c = 7.212 Å). The supercell with the QD located at the center contains a vacuum region of at least 15 Å, which is large enough to reduce the periodic interactions. All of the coordinates of atoms, including passivants, are fully relaxed until the force on each atom is no more than 0.04 eV Å⁻¹. Although the GGA underestimates the band gap, we focus specifically on the relative change of band-edges as a function of different passivating sites. As a result, the band-gap error has little effect on our fundamental conclusions.

It has been noted that in practical experiments QDs probably undergo surface reconstruction, and particularly in cases with no passivation, only a limited number of dangling bonds are left on the surface. This reconstruction changes the bulk-like electronic structures of QDs into surface electronic structures which have a great effect on the performance of QDs. For example, when the stoichiometric CdSe QDs are unpassivated, the electrons transfer from cations to anions with surface reconstruction. This reconstruction removes dangling bonds and opens the band gap, which is considered ‘self-healing’ of QDs.²¹ However, the self-healing drives sp³-like bonds between Cd and Se atoms to become surface bonds, e.g. the surface cations (Cd) form sp²-like bonds with three neighboring anionic atoms (Se) on the Cd-rich surface. These surface atomic bonds change the bulk-like electronic properties to surface electronic properties which have a high density of states at the top of the valence band. That effect becomes enhanced with decreasing size as the surface-to-volume ratio increases. Moreover, non-stoichiometric CdSe QDs may leave some chemically active bonds at the surface, which have surface states to reduce the energy gap. There are no intrinsic effects from surface reconstruction and non-stoichiometry is not discussed in the current work.

When the surface is unpassivated, inner Cd–Se bonds cause the dangling bonds of the surface Cd to be in the sp³ hybridized orbitals configuration. PH matches the orientation of the dangling bonds to conserve the sp³ configuration. The PH-terminated QDs undergo no surface reconstruction. The bonding states, due to the interaction between PH and surface atoms, locate well below the HOMO state, while the anti-bonding states are above the state of the lowest unoccupied molecular orbitals (LUMOs). With no change to the intrinsic electronic structure of band-edges, PH passivation removes the effects of surface states. The HOMO and LUMO states are mostly located within the interior of the quantum dot with negligible overlap with the states from the PH passivants. Consequently, the electronic structure of the band edges is dominated by the intrinsic chemical interaction within the interior of the QDs and is independent of the bonds between the surface atoms and the PH passivants. Theoretically, PH passivation also provides a platform to judge a different passivation strategy by comparing it with the PH passivation case.

Previous computational¹¹ and experimental²² studies confirmed that organic ligands bind preferentially to the surface Cd atoms. We focus mainly on the effect of halide ligands binding to Cd atoms. To gain insight into the effect of halide passivation, we removed some PH atoms from the surface Cd atoms at different sites and attached halide atoms (Cl or I) to the dangling bonds. The approach is similar to the replacement of H with O in the bond study of SiC.²³ Correspondingly, we employed a fully PH-terminated Cd₃₃Se₃₃ quantum dot as a reference to compare with. As the distances between the PH atoms and halide atoms are as large as 4 Å (3.5 Å for the (0001̄)^II facet case in chlorine passivation and 3.7 for the (0001̄)^II facet case in iodine passivation), the interaction between them could be reasonably neglected in the following discussion.

3 Results and discussions

Compared to the 4-coordination bonding between Cd and Se in wurtzite CdSe crystals, halide atoms prefer 3-coordination bonding with Cd due to the inactive lone pair of s² electrons of halide atoms, as demonstrated by the crystal structure of bulk CdCl₂ and CdI₂. In consideration of this, the charge of one dangling bond of one surface Cd atom is taken as a fractional charge of 1/2, while that of one halide passivant is taken to be 5/3. There is no perfect coordination combination between surface Cd absorbates and halide passivants to conform to the charge balance and coordination numbers of the Cd and halide atoms simultaneously. When one halide atom absorbs at the only dangling bond site of one 3-coordinated Cd atom, obviously the structure is unstable due to the incompletely saturated halide atom. The above consideration implies that one halide atom would interact with more than two dangling bonds from Cd to be saturated. When halide atoms attach to surface dangling bonds of QDs, two possible situations should be considered: one is that all surface Cd atoms are fully saturated and the systems maintain charge balance (i.e. one halide atom simultaneously passivating two dangling bonds of a surface Cd absorbate); the other is that one halide atom coordinates with three cadmium atoms, similar to the bonding in bulk CdCl₂ (or CdI₂) crystal to saturate the p orbitals of the halide atoms, leaving one excess electron.

The Cd positions on the surface are marked as shown in Fig. 1: A1, A2 and A3 indicate 3-coordinated Cd atoms, each with only one dangling bond to be passivated; B1 and B2 label 2-coordinated Cd atoms, each with two dangling bonds to be passivated. In general, taking chlorine passivations for example, the local structures to be passivated are classified into four categories: (i) one Cl atom passivates two dangling bonds from one 2-coordinated Cd atom (B1 or B2 sites); (ii) one Cl atom simultaneously passivates two dangling bonds from two 3-coordinated Cd atoms (A1 or A2 sites); (iii) one Cl atom passivates two dangling bonds from one 2-coordinated Cd atom and another one from one 3-coordinated Cd atom (A2 and B2 sites); (iv) one Cl atom passivates three dangling bonds from three 3-coordinated Cd atoms (A1 site) on the (0001̄) surface. In cases (i), (ii), and (iii), Cd and Cl atoms conform to the 8-valence electrons rule, which means that the QDs maintain charge balance between cationic and anionic atoms, leaving the coordination number of Cl passivants as less than 3. Case (iv) breaks the charge balance, with one excess electron from cations (Cd), but saturates the Cl passivants with coordination numbers of 3. In case (iii), there are other combinations (for example, B2 and A3 sites) which involve PH atoms bonding with Cl passivants when Cl passivants absorb at the sites. We ruled these combinations out as unstable structures, for the reason that just with surface Cd atoms, the Cl passivants can not be stable and need neighboring PH atoms to coordinate with.

Fig. 1 Calculated structures of Cl passivation on (0001)^I, (112̄0)^I, (011̄0)^II, (0001̄)^II facets of Cd₃₃Se₃₃ QDs, with the superscript of roman numbers (i.e. I, II and III) standing for the coordination number of the Cl passivants. The (011̄1)^II case stands for one Cl atom passivating Cd atoms at the A2 and B2 sites and (0001̄)^III stands for one Cl atom passivating three Cd atoms at the A1 site. Cd, Se, Cl and PH atoms are colored grey, yellow, green and blue, respectively. The (0001̄)^II and (0001̄)^III cases are viewed from the bottom, while the other 4 cases are viewed from the top.

Based on the relaxed structures, the configuration of the above four cases can be divided into two categories: (i) for mono-dentate Cl passivants, i.e. one Cl passivant coordinating to one 2-coordinated Cd atom; the addition of Cl leads to no obvious surface reconstruction of the wurtzite structure of QDs (see discussion in Section 3.1); (ii) for multi-dentate Cl passivants, i.e. one Cl passivant coordinating to more than one surface Cd atom; the addition of Cl affects the bond length of a few Cd–Cd pairs and results in surface reconstruction to some extent (see discussion in Section 3.1).

3.1 Chlorine passivation with nearly valid eight-valence electrons rule

When one chlorine atom coordinates with one 2-coordinated cadmium atom (B1 or B2 site in Fig. 1) on the (0001)^I or (112̄0)^I facet, the passivation maintains the surface of the QDs in the same configuration as bulk wurtzite CdSe. Due to the dominant ionic bonding character, the interaction between the Cl passivants and the surface Cd atoms is electrostatic interaction which passivates dangling bonds. The electrostatic interaction pushes the surface states of the dangling bonds of the cations down below the HOMO state, but not as deep as those in the PH-terminated cases. Additionally, due to the electrostatic interactions, the bond lengths between Cl passivants and surface Cd absorbates become important (see Table 1). We find that the bond lengths between surface Cd absorbates and Cl passivants get longer as the coordination number of the Cl passivants increases. The energy gap is composed of the bonding state and anti-bonding state of Cd–Se in the PH passivation case. However, in Cl passivation cases, the bonding state from surface Cd absorbates and incompletely saturated Cl atoms (except for the (0001̄)^III case) slightly perturbs the HOMO states of the QDs, depending on the coordination number of the Cl passivants. In incompletely saturated Cl passivation, the energy level of the bonding state of Cd–Cl, which depends on the bond length of Cd–Cl, is higher than those of Cd–Se and occupies the HOMO states of the QDs. Also, the energy gap is bigger with shorter bond lengths of Cd–Cl (see Table 1). Details will be discussed below.

Table 1 Calculated energy gaps and bond lengths of Cd₃₃Se₃₃ quantum dots with PH passivation, Cl passivation and I passivation on topical facets. The energy gap is the energy difference between the HOMO state and LUMO state. Bond length is estimated based on the distance between the passivants and the surface Cd absorbates

		Chlorine passivation						Iodine passivation
	PH	(0001)^I	(112̄0)^I	(011̄0)^II	(0001̄)^II	(011̄1)^II	(0001̄)^III	(0001)^I	(0001̄)^II	(0001̄)^III
Energy gap (eV)	2.58	2.53	2.49	2.43	2.47	2.40	2.50	2.06	2.33	2.42
Bond length (Å)	1.88	2.35	2.37	2.65	2.65	2.57/2.70	2.83	2.66	2.94	3.14

---

Generally speaking, all passivation cases have LUMOs similar to PH-terminated QDs. With negligible overlap with Cl passivants, the LUMO states are mainly located at the interior Cd atoms (see Fig. 2). The energy levels of the anti-bonding state of Cd–Cl are higher than those of interior Cd–Se. Passivated surface Cd atoms and incompletely saturated Cl passivants do not lead to surface states close to the LUMO states. As a result, the LUMO states of the QDs are composed of intrinsically anti-bonding states of Cd–Se, and the incompletely saturated Cl passivants have a negligible effect on the LUMO states.

Fig. 2 HOMO (left column) and LUMO (right column) of Cl-passivated Cd₃₃Se₃₃ QDs at Γ point.

When Cl passivants are mono-dentate, the HOMO states of the QDs with Cl passivation on the (0001)^I facet and the (112̄0)^I facet are composed of the incompletely saturated Cl 3p orbitals (see Fig. 2), partially mixing with neighboring Se 4p orbitals. The states of Cl 3p orbitals are above those of the intrinsic Cd–Se of the QDs. Although Cl passivants remove the surface states of the dangling bonds of surface Cd atoms, the incompletely saturated Cl passivants lead to new surface states. The relatively high density of Cl 3p states actually leaves the QDs unstable, and the QDs prefer to lower the energy level of the Cl p orbitals through surface structure relaxation. As a result, the (0001)^I facet and (112̄0)^I facet are the last possible sites on to which Cl passivants absorb.

In the case of (0001̄)^II (or (011̄0)^II) facet passivation, the bond lengths between Cl and 3-coordinated Cd atoms are shorter than the Cd–Cl distance in bulk CdCl₂ (2.70 Å). The HOMO states are also contributed to by the Se 4p and Cl 3p states. Although, the (0001̄)^II and (011̄0)^II cases reduce the contribution from Cl 3p states to the HOMO states, the mixture breaks the symmetry of the distribution of HOMO states in the PH-terminated case. Moreover, the spatial wavefunction distribution of the top HOMOs shows that those wavefunctions have a significant contribution from the surface Cl passivants of the QDs. Compared with the PH-terminated case, those 2-coordinated Cl atoms significantly affect the electronic structure of band edges. Consequently, the HOMO states of the (0001̄)^II and (011̄0)^II cases are a mixture of Se 4p and Cl 3p states, compared with those in the PH-terminated case which are Se 4p states. In consideration of this, Cl passivants in bidentate cases (i.e. (0001̄)^II and (011̄0)^II cases) remove surface states of surface Cd absorbates but perturb the HOMO states of QDs. Compared with the mono-dentate case of Cl passivation, bidentate cases lower the energy level of Cl passivants, which is still higher than that in the tridentate case (see discussion in Section 3.2).

Fig. 3 shows the schematic bonding picture of charge balance cases. The above discussed Cl passivation cases, conforming to the 8-valence electrons rule, remove the dangling-bond-derived states. The LUMO states of the Cl passivation cases can be considered to be the same as those in the PH-terminated case, which means that Cl passivants remove the dangling-bond-derived states and cause negligible perturbation to the LUMO states of the QDs. However, the passivation effect of Cl passivation on band edges is not as strong as that in PH passivation, and the contribution of Cl passivants to the HOMO states can still be clearly observed, depending on the coordination number of the Cl passivants. Additionally, the energy levels of the bonding states of the coupling surface Cd and incompletely saturated Cl passivants are higher than those of the bonding state between Cd and Se atoms. As shown in Fig. 2, for all Cl passivation cases, the incompletely saturated Cl passivants, i.e. the coordination numbers of the Cl passivants less than three, cause great perturbation to the HOMO states. Though Cl passivants can passivate, to a great extent, the dangling bonds of surface Cd absorbates in all cases, the p orbitals of Cl atoms need three neighboring Cd atoms to be saturated. However, the charge balance consideration (i.e. all atoms conforming to the 8-valence electrons rule) leaves the coordination number of Cl passivants at less than three. Also, a mixture of Cl 3p states and Se 4p states composes the HOMO states of the QDs, and the coordination number of the Cl passivants influences the wavefunction distribution of the HOMO states by changing the saturation extent of the Cl passivants. As a result, the energy levels of the bonding states of Cd–Cl in charge balance Cl passivations are not definitely below that of the intrinsic bonding state of Cd–Se but can be considered as a function of the coordination number of the Cl passivants. With the coordination number of the Cl passivants increasing, the energy level of the bonding states of Cd–Cl gets deeper. The system in which one chlorine atom coordinates with two 3-coordinated Cd atoms is more stable than the system in which one chlorine atom coordinates with only one 2-coordinated Cd atom, but the most stable case should correspond to the Cl passivants interacting with 3-coordinated Cd atoms. The HOMO states with incompletely saturated Cl p orbitals are expected to have a negative effect on the performance of QD-based devices.

Fig. 3 Schematic bonding picture of PH passivation and Cl passivation.

3.2 Chlorine passivation with 3-coordinated Cd atoms

Effective Cl passivation depends on the distance between the Cl passivants and surface Cd absorbates. Only in the case of (0001̄) facet passivation, one chlorine atom can directly bond to three 3-coordinated Cd atoms (A1 site) with a distance of 2.83 Å, which is 4.8% larger than the calculated Cd–Cl bond length (2.70 Å) in bulk CdCl₂. In such cases, the HOMO–LUMO gap is 2.50 eV, which is 0.03 eV larger than the gap for cases in which one chlorine atom is passivating two 3-coordinated Cd atoms on the (0001̄)^II facet (see Table 1). Though the charge balance of QDs is broken, the chemical bonding from Cl–Cd interactions acts as ‘ligand potential’ to passivate the dangling bonds of three Cd atoms and remove surface states, leaving one excess electron. As shown in Fig. 4(a), although the excess charge pushes the Fermi energy up close to the LUMO states as in n-type doping, the ‘ligand potential’ pushes surface states derived from the dangling bonds of three surface Cd atoms below the HOMO states and the tridentate Cl passivant is saturated. Unlike dangling bonds, the excess electron shifts the Fermi energy level but does not inject new surface states into the band gap. As a result, one chlorine atom passivates three 3-coordinated surface Cd absorbates.

Fig. 4 Density of states and chlorine passivations on the (0001̄) surface. (a) The total density of states (DOS) of Cd₃₃Se₃₃ QDs with the (0001̄) surface passivated by PH; one Cl bonding to two 3-coordinated Cd atoms in the (0001̄)^II case, and one Cl bonding to three 3-coordinated Cd atoms in the (0001̄)^III case, respectively. The zero of energy is aligned with the top of the valence band and the dashed line indicates the Fermi levels for each case. (b) and (c) are the wavefunction distributions of the HOMO and LUMO states for the (0001̄)^III case.

Due to the large difference in the electronegativities between Cl and Cd, Cl 3p and Cd 5s have weak p–s mixing in the upper valence band in CdCl₂ crystals. Consequently, the Cl 3p occupied states form the top of the valence band, which is almost 1 eV deeper than those of PH-terminated Cd₃₃Se₃₃ QDs, and a mixture of Cd 5s–Cl 3p unoccupied states form the bottom of the conduction band.²⁴ As shown in Fig 4(b) and (c), in the (0001̄)^III case, the HOMO wavefunction of the QD is located at Se sites and has no overlap with Cl passivants, and the LUMO wavefunction slightly overlaps with Cl passivants. The 3-coordinated Cd atoms, as in the case of bulk CdCl₂ crystals, saturate Cl passivants and lower the energy level of Cl 3p states below HOMO state, so that the HOMO states are located at the interior Se atoms of the QDs. Energetically, the passivation of one Cl passivant bonding with three 3-coordinated Cd atoms is more stable than the passivations involving one Cl passivant bonding with fewer Cd atoms because the latter cases incompletely saturate Cl which leads to higher energy states of the Cl passivants.

3.3 Iodine passivation on QDs

The chemical bonding between Cl passivants and surface Cd atoms could remove surface states derived from the dangling bonds of surface cations, even in the cases in which the charge balance is broken (see Section 3.2). However, because of the high electronegativity of Cl, the Cl passivant needs three Cd atoms to form a tridentate ligand to be well saturated. With only a few sites on the surface of the QD providing three sufficient cations to coordinate, the vast majority of Cl passivants tend to interact with cations from the surfaces of neighboring QDs to be saturated. A recent experiment¹³ reported that CdSe QDs with Cl passivants tend to aggregate but those with Br or I passivants undergo no obvious changes in size and show no sign of aggregation, which seems partly consistent with our conclusions. Fig. 5 plots the electronic structures of QDs passivated by I atoms on (0001)^I, (0001̄)^II, and (0001̄)^III facets, respectively. I passivants could remove the surface dangling states and broaden the energy gap (see Fig. 5a). Meanwhile, the I passivant with a mono-dentate ligand also leads to new surface states above the bonding states of Cd–Se in the (0001)^I case. In the case of bidentate I passivants, the chemical bonding from I–Cd lowers the energy level of the I passivants and the HOMO states are composed of a mixture of I 5p and Se 4p in the (0001̄)^II case. However, even with I passivants dangling with three Cd coordinations, the HOMO states are still composed of a mixture of I 5p and Se 4p in the (0001̄)^III case (see Fig. 5b). This is partly different from the case of chlorine passivation.

Fig. 5 Density of states and iodine passivation on the (0001) and (0001̄) surfaces. (a) The total density of states (DOS) of Cd₃₃Se₃₃ QDs passivated by PH, one I bonding to one 2-coordinated Cd atom in the (0001)^I case, one I bonding to two 3-coordinated Cd atoms in the (0001̄)^II case, and one I bonding to three 3-coordinated Cd atoms in the (0001̄)^III case, respectively. The zero of energy is aligned with the top of the valence band and the dashed line indicates the Fermi levels for each case. (b) and (c) are the wavefunction distributions of the HOMO (middle panel) and LUMO states (bottom panel) for the three cases.

Due to the difference in electronegativity between Cl and I, and the varying bond lengths, the chemical bonding between I–Cd is expected to be weaker than that of Cl–Cd. Consequently, the energy levels of the bonding states of I–Cd are not as deep as those of Cl–Cd. The HOMO states in the (0001̄)^III case, which are composed of I 5p and Se 4p, demonstrate that the coordination numbers of the I passivants have less of an effect on the HOMO states in iodine passivation than in chlorine passivation. In chlorine passivation, the energy levels of the bonding states of Cd and 3-coordinated Cl atoms fall deeper in energy than those of the bonding states of Cd and Se atoms within QDs, but the incompletely saturated Cl passivants have relatively high energy levels, which are comparable with the energy levels of the bonding states of Cd–Se in QDs (see Fig. 2). To lower the activity of the incompletely saturated Cl passivants, the Cl passivants tend to interact with nearby or neighboring cations, leading to the aggregation of the QDs. However, in iodine passivation, due to the decrease in the electronegativity, the energy levels of Cd and 3-coordinated I atoms are intrinsically comparable with the energy levels of the bonding states of Cd–Se in the QDs. Also, the coordination number of I passivants has less of an effect on the saturation of I passivants than of Cl passivants, because the passivation of one I passivant to only two 3-coordinated Cd atoms has already been stable in iodine passivation cases. Therefore, QDs with 2-coordinated I passivants have weak interactions with cations from neighboring QDs, leading to no obvious aggregation and change in size.

In the PH-terminated case, PH atoms, exactly matching the direction and fractional charge of dangling bonds, perfectly passivate surface cations. The bonding states and anti-bonding states from PH and surface atoms can be pushed down deep into the valence band and up into the conduction band, respectively. Organic ligands usually passivate surface atoms through surface reconstruction, which leads to charge transfer from cationic to anionic atoms. The binding between organic ligands and 2-coordinated Cd atoms is always stronger than that between organic ligands and 3-coordinated Cd atoms. In addition, the effect of organic ligands depends on the binding capacity of the ligand to the QD as well as steric inter-ligand interactions.¹² However, after removing the surface states derived from the dangling bonds of cations of QDs, the effect of Cl passivation depends mainly on the coordination number of the Cl passivants. The incompletely saturated Cl passivants may affect the electronic structure of band edges, while completely saturated Cl passivants have little effect on edge states. For small QDs, the different coordination environment of Cl passivants may lead to different distributions of the HOMO states, which potentially affects the efficiency of QD-based devices. Compared with organic ligand passivation based on binding capacity and complex surface reconstruction, halide passivants perfectly present better passivation due to electrostatic interaction.⁵ Moreover, due to the different preferred absorbate sites, organic ligands prefer to absorb to one 2-coordinated cations and halide passivants prefer to bind with more than one 3-coordinated cation, a hybrid system of organic and halide ligands can improve the efficiency of solar cells.⁶ Steric locations to which halide passivants absorb may compensate for the organic ligand-uncovered surface and improve the coverage and thus the passivation of QDs.

4 Conclusion

In summary, we have investigated the electronic structures of QDs passivated by Cl and I passivants based on DFT calculations. Our calculations demonstrate that, unlike the perfect coordination number match in PH passivation, Cl passivants need to be 3-coordinated to be completely saturated, while the electrostatic interaction between Cl and Cd atoms can passivate dangling bonds of surface Cd atoms, even in the case of the charge balance being broken. The Cl passivation may lead to some changes in the electronic structure of QDs, depending on the coordination number of the Cl passivants. With an increasing Cl passivant coordination number, the HOMO state has less contribution from the Cl passivants and the QDs become energetically more stable, and the energy levels of Cl passivants are lowered. Incompletely saturated Cl passivants may interact with cations of neighboring QDs and, thus, result in aggregation. With electronegativity decreasing from Cl to I, the states of I passivants 5p compose the HOMO states of passivated QDs and have an effect on the electronic structure. Also, 2-coordinated I passivants are energetically stable, resulting in no obvious aggregation of QDs. Our analysis seems helpful to explain the passivation of organic and inorganic ligands that could potentially improve the coverage and the efficiency of QDs, besides taking steric effects into consideration.

Acknowledgements

This work is supported by National Natural Science Foundation of China Grants (11234012 and 51121064).

References

1. A. P. Alivisatos, Science, 1996, 271(5251), 933–937.
2. J. Jasieniak, M. Califano and S. E. Watkins, ACS Nano, 2011, 5(7), 5888–5902.
3. G. D. Scholes, Adv. Funct. Mater., 2008, 18(8), 1157–1172.
4. P. Guyot-Sionnest, B. Wehrenberg and D. Yu, J. Chem. Phys., 2005, 123(7), 074709.
5. J. Tang, K. W. Kemp, S. Hoogland, K. S. Jeong, H. Liu, L. Levina, M. Furukawa, X. Wang, R. Debnath, D. Cha, K. W. Chou, A. Fischer, A. Amassian, J. B. Asbury and E. H. Sargent, Nat. Mater., 2011, 10, 765–771.
6. A. H. Ip, S. M. Thon, S. Hoogland, O. Voznyy, D. Zhitomirsky, R. Debnath, L. Levina, L. R. Rollny, G. H. Carey, A. Fischer, K. W. Kemp, I. J. Kramer, Z. Ning, A. J. Labelle, K. W. Chou, A. Amassian and E. H. Sargent, Nat. Nanotechnol., 2012, 7(9), 577–582.
7. H. Liu, J. Tang, I. J. Kramer, R. Debnath, G. I. Koleilat, X. Wang, A. Fisher, R. Li, L. Brzozowski, L. Levina and E. H. Sargent, Adv. Mater., 2011, 23, 3832–3837.
8. M. C. Beard, A. G. Midgett, M. Law, O. E. Semonin, R. J. Ellingson and A. J. Nozik, Nano Lett., 2009, 9(2), 836–845.
9. D. V. Talapin, A. L. Rogach, A. Kornowski, M. Haase and H. Weller, Nano Lett., 2001, 1(4), 207–211.
10. Y. Chen, J. Vela, H. Htoon, J. L. Casson, D. J. Werder, D. A. Bussian, V. L. Klimov and J. A. Hollingsworth, J. Am. Chem. Soc., 2008, 130(15), 5026–5027.
11. A. Puzder, A. J. Williamson, N. Zaitseva, G. Galli, L. Manna and A. P. Alivisatos, Nano Lett., 2004, 4(12), 2361–2365.
12. S. A. Fischer, A. M. Crotty, S. V. Kilina, S. A. Ivanov and S. Tretiak, Nanoscale, 2012, 4(3), 904–914.
13. G. D. Niu, L. D. Wang, R. Gao, W. Z. Li, X. D. Guo, H. P. Dong and Y. Qiu, Phys. Chem. Chem. Phys., 2013, 15, 19595–19600.
14. S. Kilina, S. Ivanov and S. Tretiak, J. Am. Chem. Soc., 2009, 131(22), 7717–7726.
15. M. S. de la Fuente, R. S. Sánchez, V. González-Pedro, P. P. Boix, S. G. Mhaisalkar, M. E. Rincón, J. Bisquert and I. Mora-Seró, J. Phys. Chem. Lett., 2013, 4(9), 1519–1525.
16. Y. Q. Zhang and X. A. Cao, Appl. Phys. Lett., 2011, 99(2), 023106.
17. H. T. Zhang, B. Hu, L. F. Sun, R. Hovden, F. W. Wise, D. A. Muller and R. D. Robinson, Nano Lett., 2011, 11(12), 5356–5361.
18. X. Y. Huang, E. Lindgren and J. R. Chelikowsky, Phys. Rev. B: Condens. Matter Mater. Phys., 2005, 71(16), 165328.
19. H. X. Deng, S. S. Li, J. B. Li and S. H. Wei, Phys. Rev. B: Condens. Matter Mater. Phys., 2012, 85(19), 195328.
20. A. Kasuya, R. Sivamohan, Y. A. Barnakov, I. M. Dmitruk, T. Nirasawa, V. R. Romanyuk, V. Kumar, S. V. Mamykin, K. Tohji, B. Jeyadevan, K. Shinoda, T. Kudo, O. Terasaki, Z. Liu, R. V. Belosludov, V. Sundararajan and Y. Kawazoe, Nat. Mater., 2004, 3(2), 99–102.
21. A. Puzder, A. J. Williamson, F. Gygi and G. Galli, Phys. Rev. Lett., 2004, 92(21), 217401.
22. E. L. Rosen, R. Buonsanti, A. Llordes, A. M. Sawvel, D. J. Milliron and B. A. Helms, Angew. Chem., Int. Ed., 2012, 51, 684–689.
23. A. Trejo, M. Calvino, E. Ramos and M. Cruz-Irisson, Nanoscale Res. Lett., 2012, 7(1), 471.
24. I. N. Demchenko, M. Chernyshova, W. C. Stolte, D. T. Speaks and A. Derkachova, Mater. Chem. Phys., 2012, 135(2–3), 1036–1043.

---