Electronic Structure and Surface Properties of Copper Thiocyanate: A Promising Hole Transport Material for Organic Photovoltaic Cells

Considering the significance of hexagonal copper thiocyanate (β-CuSCN) in several optoelectronic technologies and applications, it is essential to investigate its electronic structure and surface properties. Herein, we have employed density functional theory (DFT) calculations to characterise the band structure, density of states, and the energy-dependent X-ray photoelectron (XPS) valence band spectra at variable excitation energies of β-CuSCN. The surface properties in the absence and presence of dimethyl sulfoxide (DMSO), a solvent additive for improving perovskite solar cells’ power conversion efficiency, have also been systematically characterised. β-CuSCN is shown to be an indirect band gap material (Eg = 3.68 eV) with the valence band edge demonstrated to change from being dominated by Cu-3d at soft X-ray ionisation photon energies to Cu-3p at hard X-ray ionisation photon energies. The adsorption energy of dimethyl sulfoxide (DMSO) on the (100) and (110) β-CuSCN surfaces is calculated at −1.12 and −0.91 eV, respectively. The presence of DMSO on the surface is shown to have a stabilisation effect, lowering the surface energy and tuning the work function of the β-CuSCN surfaces, which is desirable for organic solar cells to achieve high power conversion efficiencies.


Introduction
Halide perovskite solar cells (PSCs) are attractive, cost-effective, and earth-abundant candidates for solar cell development [1][2][3]. Because of the distinctive physical and optoelectronic properties, the organic-inorganic halide perovskites with typical ABX 3 formula have attained certified overall conversion efficiencies exceeding 22% [4][5][6], making these materials suitable candidates to replace commonly employed Si-based solar cells. Perovskite solar cell device architecture consists of three important layers: the active photoabsorber (halide perovskite) layer, a hole-transporting layer (HTL), and an electron-transporting layer (ETL), to extract holes and electrons, respectively.

Materials and Methods
The DFT calculations were performed in the VASP-Vienna Ab initio Simulation Package [37][38][39]. The interactions between the valence and core electrons are described using the projected augmented wave (PAW) method [40]. The Perdew-Burke-Ernzerhof (PBE) exchange-correlation function is used for geometry optimisations [41]. To accurately predict the electronic band gap and density of states (DOS) features of CuSCN, the screened hybrid functional HSE06 [42] with an exchange value of 25% was used. A kinetic energy cut-off of 600 eV and Monkhorst-Pack [43] K-points mesh of 7 × 5 × 3 converge the total energy of β-CuSCN to within 10 −6 eV, and the residual forces on all relaxed atoms reached 10 −3 eVÅ −1 .
The METADISE code [44] was used to create the (100) and (110) β-CuSCN surfaces. The stability of the surfaces is quantified by calculating the surface energy (γ r ) of the relaxed naked surfaces as: where E relaxed sur f is the total energy of the relaxed surface, nE bulk is the energy of an equal number (n) of the bulk β-CuSCN atoms, and A is the surface area. To avoid interactions between periodic slabs, a 20 Å vacuum is added to the c-direction. To ascertain whether the adsorption of DMSO has a stabilisation effect on the surface, we have also computed surface energy of the one-side (1 × A) DMSO-covered β-CuSCN surface, calculated as: where E relaxed sur f +DMSO is the total energy of the DMSO-covered surface and E DMSO is the total energy of the free DMSO molecule. The adsorption energy (E ads ) of DMSO on β-CuSCN surfaces is calculated as: Negative or positive adsorption energy denotes an exothermic or endothermic adsorption process, respectively. The averaged electrostatic potential along the c-direction of the (100) and (110) β-CuSCN surfaces was determined using the MacroDensity package [45,46]. The work function (Φ), the minimum energy required to eject an electron from a solid into vacuum, is determined as: is the vacuum level and E F is the Fermi level. Dipole corrections were accounted for in all surface calculations to ensure accurate determination of the vacuum level's potential [47]. The XPS valence band spectra were simulated using the GALORE code [48], which used the Gelius model [49,50] to apply weightings to the atomic projected density of states (PDOS) based on the photoionisation cross-sections formulated by Scofield [51,52].

Electronic Properties of β-CuSCN
β-CuSCN has an indirect band gap of 3.68 eV, as evident in the band structure ( Figure 2a). The hybrid HSE06 functional predicted band gap compares closely with commonly reported experimental values in the range of 3.6-3.9 eV [56][57][58][59] and previous theoretical calculations [36,54,55]. The partial density of states (PDOS) of β-CuSCN is shown in Figure 2b. The valence band maximum (VBM) is dominated by Cu-3d and S-2p states, whereas conduction band minimum (CBM) is composed mainly of C-2p and N-2p, with a small contribution from C-3d. To gain insight into the superior hole-transport property of β-CuSCN, we have computed the effective masses of holes (m * h ) and electrons (m * e ) by fitting a quadratic polynomial to the energy versus reciprocal lattice As shown in Table 2, the predicted m * h (at the valence band maximum) and m * e (at the conduction band minimum) in the Brillouin zone's selected directions are generally very small, which point to their superior transport. The smallest effective masses of holes (m * h ) appear in the direction from Γ to A (0.0037), whereas the highest appears in the Γ−M (0.1756) direction. These suggest that photogenerated holes in β-CuSCN should be easiest to transport along the Γ−A direction and least mobile in the Γ−M direction.  The energy-dependent XPS valence band spectra of β-CuSCN at variable excitation energies (Al Kα1 (1486 eV), 4068 eV, and 8133 eV), obtained by applying weightings to the DFT-PDOS using the Scofield [51] formulated photoionisation cross-sections are shown in Figure 3. The chosen photon energies correspond to moving from soft (1486 eV) to intermediate (4068 eV) to hard (8133 eV) X-ray ionisation photon energy (hν). Five spectral features are assigned: I for the main valence band region (onset from 0.0 eV and extends to 3.4 eV), II, III, IV, and V for the broad satellite features located around 4, 9, 17, and 21 eV binding energies, respectively. The intensity of I relative to IV and V decreases continuously for increasing photon energy, which resulted in modifying the main valence band spectrum's shape and increasing the broad satellite features' relative intensity. The changes in the valence band spectral features may be attributed to the photon energy dependence of the photoionisation cross-section to the photon energy dependence of the photoionisation cross-section of the atomic s, p, and d orbitals that contributes to the valence band. Cu-3d states dominate the main valence band region (I) for β-CuSCN at lower photon energies, owing to their higher photoionisation cross-section (Figure 3d). The Cu-3d photoionisation cross-section decreases much more rapidly than Cu-3p with increasing photon energy, so its contribution becomes less significant (Figure 3a-c). This resulted in the main valence band (I and II) switching from being dominated by Cu-3d to Cu-3p. The satellite broad-feature IV can be assigned to the main contribution from S-3s, whereas the broad-feature V is assigned to contributions from N-2s. The simulated main valance band features at soft X-ray ionisation photon energy (1.486 keV) shows good agreement with the XPS valence band spectrum of CuSCN thin films reported by Jaffe et al. [36], which demonstrate that Cu-3d states dominate the main valence band region for β-CuSCN at lower photon energies. These results are also consistent with the work of Wijeyasinghe et al. [57], who showed that Cu-3d peaks dominate the main valence band.

Surface Properties
Here, we have characterised the structure, composition, relative stabilities, and work function of the two most stable β-CuSCN (100) and (110) surfaces. Both surfaces have two non-dipolar terminations (denoted as termination T1 and T2), both of which are characterised to determine the most stable one. Shown in Figures 4 and 5 are the relaxed structures of the two terminations of the (100) and (110) β-CuSCN surfaces. Termination T1 (T2) of the (100) surface has a calculated surface energy of 0.57 (0.76 Jm −2 ), indicating that the termination T1 is more stable than termination T2. Similarly, we predict the surface energy of T1 (T2) of the (110) surface to be 0.43 (0.62 Jm −2 ). Similarly, results were reported by Chen et al. [54], who reported surface energy of 0.74 and 0.63 Jm −2 for the most stable terminations of the (110) and (100) surfaces, respectively.
Next, we investigated the adsorption characteristics of dimethyl sulfoxide (DMSO) on the most T1 termination of the β-CuSCN (100) and (110) surfaces to ascertain adsorption effects on the surface stability and work function. These insights are essential, considering that residues of DMSO solvent that is typically used to fabricate solution-processable halide perovskite materials [60] may adsorb onto the HTL β-CuSCN surfaces, altering the electronic properties. The DMSO molecule is introduced to the (100) and (110) surfaces in different orientations and at various sites to predict the most stable adsorption geometries, as shown in Figure 6. At both surfaces, the DMSO interacts with the Cu sites through the O atom, releasing an adsorption of −1.12 and −0.91 eV at the (100) and (110) surfaces, respectively. The interacting Cu-O bond distance at the (100) surface is calculated at 2.126 Å, whereas at the (110) surface, it is 2.191 Å (Table 3). Our predicted adsorption structures and energetics of DMSO at the β-CuSCN (100) and (110) surfaces are in good agreement with those reported by Zhang and Chen et al. [61], who show that the solvent DMSO molecule exhibits reasonable interactions with the CuSCN surfaces, predicting adsorption energy of 1.28 and 0.84 eV at the (001) and (110) surfaces, respectively.    The adsorption of the DMSO on the (100) and (110) surfaces is demonstrated to be characterised by redistribution of charge density in the surface-DMSO systems, as obtained from analysis of the differential charge density (∆ρ) iso-surface contours (Figure 6c,d).
where ρ sur f +DMSO , ρ sur f , and ρ DMSO are the charge density of the total surf-DMSO complex, the naked β-CuSCN surface, and the free DMSO molecule as in the relaxed adsorbed configuration. We observe charge density accumulation between the Cu-O bonds, suggesting chemisorption. Bader charge analyses reveal that the net charge transfer from the β-CuSCN surfaces to the DMSO molecule is minimal, calculated to be 0.08 and 0.05 e − at the (100) and (110) surfaces, respectively (Table 3).
Owing to the small charge transfer to the DMSO molecule upon adsorption, the interatomic bond distances and angles changed marginally, relative to the gas-phase parameters.
The DMSO-covered surfaces are predicted to have smaller surface energies compared to the naked surfaces, suggesting that the DMSO adsorption has a stabilisation effect on the (100) and (110) β-CuSCN surfaces. The DMSO-covered β-CuSCN(100) surface energy is calculated at 0.49 Jm -2 , compared to 0.57 Jm −2 for the naked surface. Similarly, the DMSO-covered β-CuSCN (110) surface has a surface energy of 0.39 Jm −2 , compared to 0.43 Jm −2 for the naked surface. The adsorption of DMSO on the (100) and (110) β-CuSCN surfaces modifies their work function (Φ). Hole transport materials with a high degree of tunable work functions that permit proper energy alignment with the donor materials' deep valence bands are desirable for high-efficiency organic solar cells [62]. The Φ of the naked CuSCN (100) surface is computed at 5.46 eV vs. vacuum, as shown in Figure 7a, compared to 5.23 eV for the DMSO-covered (100) surface (Figure 7b). For the CuSCN (110) surface, the Φ naked and DMSO-covered surfaces are predicted at 5.27 and 5.05 eV, respectively (Figure 7c,d). The reduction in the work function upon DMSO adsorption may be rationalised by considering the fact that the DMSO adsorption is characterised by electron density rearrangement, which smoothens the surface electric charge distribution according to the Smoluchowski effect [63,64]. Similar results have been reported for other organic functionalised surfaces [65][66][67][68]. These calculated work functions compare closely with experimental measurements, which report the Φ of CuSCN in the range of 5.0-5.4 eV [69][70][71][72]. Wijesekara et al. [70] reported the work function of CuSCN thick film to be 4.82 eV. Dzikri et al. [71] and Kim et al. [72] reported the work functions of 5.53 and 5.4 eV, respectively. The predicted work function of CuSCN also compares with those of other commonly used hole transport layer materials: Spiro-OMeTAD = 4.9, PEDOT:PSS = 5.1 eV, NiO x = 5.4 eV, V 2 O 5 = 5.4 eV, MoO x = 5.3 eV, WO 3 = 5.35 eV, and GO = 4.9 eV [73,74]. The close match between the work function of CuSCN and the highest occupied molecular orbital level of most perovskite materials (e.g., CH 3 NH 3 PbI 3 = -5.4 eV) ensures good alignment of energy levels, which will give rise to improvements in device efficiency.

Summary and Conclusions
In summary, we have comprehensively characterised the electronic structure, energy-dependent XPS valence band spectra at variable excitation energies, and the surface properties of β-CuSCN in the absence and presence of dimethyl sulfoxide (DMSO), using first principles density functional theory calculations. The β-CuSCN is shown to possess an indirect large band gap (E g = 3.68 eV) with the valence band edge demonstrated to change from being dominated by Cu-3d at soft photon energies to Cu-3p at hard X-ray ionisation photon energies. The adsorption of DMSO on the (100) and (110) β-CuSCN surfaces was demonstrated to have a stabilisation effect, lowering the surface energy. The adsorption was shown to be characterised by electron redistribution in the DMSO-CuSCN systems, leading to accumulation within the interacting Cu-O bonding regions. The presence of the DMSO was shown to tune the work function of the CuSCN surfaces, which is desirable for hole transport layer materials to achieve improved solar device characteristics and performance.