Panagiotis
Dallas
*ab,
Vasileios K.
Tzitzios
*b,
Lida
Givalou
b,
Polychronis
Tsipas
bc,
Georgia
Basina
ab,
Elias
Sakellis
bd,
Nikos
Boukos
b and
Thomas
Stergiopoulos
*b
aTheoretical and Physical Chemistry Institute, National Hellenic Research Foundation, 48 Vassileos Constantinou Avenue, Athens 11635, Greece. E-mail: pdallas@eie.gr
bInstitute of Nanoscience and Nanotechnology, NCSR Demokritos, 15341, Athens, Greece. E-mail: p.dallas@inn.demokritos.gr; v.tzitzios@inn.demokritos.gr; t.stergiopoulos@inn.demokritos.gr
cNational Institute of Materials Physics, Atomistilor, 405A, Magurele, Romania
dDepartment of Physics, National and Kapodistrian University of Athens, Greece
First published on 12th March 2025
Ag8SnS6 (ATS) nanoparticles, with a band gap of 1.35 eV, which is located exactly at the Schockley–Queisser optimal value for a single-junction solar cell, were utilized as a photoabsorber component in solid state photovoltaic devices. The as-made particles were capped with long aliphatic chains of oleic acid and oleylamine. After surface functionalization of the shorter and extremely basic formamidinium cations, an increase of the absorption coefficient throughout the visible spectrum range was observed. The ligand exchange led also to a slight increase of the band gap, by a value of 0.05 eV. XRD, XPS, UPS, diffuse reflectance, TEM and EDX characterization studies revealed the structure of the nanoparticles and finally proof-of-concept thin film solar cells were fabricated. A maximum photoconversion efficiency of 0.22% was achieved for the as-made particles.
Inorganic perovskites such as BaZrS3 are known to demand high temperature solid state procedures to be synthesized. However, recent reports have demonstrated their solution-based synthesis towards colloidal nanoparticles, which is expected to provide new pathways for their applications in photovoltaics.8 On the other hand, AgBiS2 can be easily synthesized in solution with the ligands that are decorating its surface appearing to influence the photovoltaic performance.9,10 AgBiS2 recently has achieved a record efficiency of 10.2% when submicron grain-thin films through a vapor assisted solution process were applied,11 while cation-disorder-engineered colloidal nanocrystals reported a certified PCE value of 8.85%.12,13
An efficient solar absorber material for practical applications needs an energy gap, Eg, near 1.34 eV to produce maximal output power.14 However, only a small number of binary semiconductors satisfies this requirement and limits the panel of materials suitable for solar absorbers in photovoltaics. However, an advantage of the ternary chalcogenides semiconductors is that the band gap can be tuned by varying the ratios of the constituent elements. To that end, the I–III/IV–VI (Cu- and Ag-based) ternary metal sulfides, have been extensively studied as candidates for applications in photovoltaics as well as light emitting diodes.15
Ag8SnS6 with the canfieldite crystal structure has an ideal Eg of 1.3–1.5 eV, and high absorption coefficients of α ∼ 104 cm−1 in the visible range.16 Ag, Sn and S are non-toxic, relatively low-cost elements and environmentally friendly. With respect to the above, Ag8SnS6 nanocrystals have been utilized as counter electrodes in dye-sensitized solar cells,17 photocatalytic dye degradation,18 and photoelectrochemical salt-water splitting.19 Recently, Ag8SnS6 photovoltaic devices have been fabricated with an efficiency of 0.25%.20 Zhu et al. employed spiro-OMeTAD as hole transport layer and mesoporous TiO2 as electron transport layer to fabricate the thin film solar cells.
Here, we report a simple, straightforward, solution-based preparation of Ag8SnS6 nanocrystals and investigation of their photovoltaic properties. We investigated ATS nanoparticles capped with two different ligands. Specifically, we compared nanoparticles functionalized with the bulky, long, aliphatic chains of oleic acid and oleylamine with nanoparticles capped with the short formamidinium (FA+) cation. In both cases, photovoltaic devices were fabricated and a strong increase in the absorption coefficient was demonstrated after functionalization with FA+.
Furthermore, EDX mapping identified the presence and the distribution of the Ag, S and Sn elements in the crystallites of both Pr–Ag8SnS6 and FA–Ag8SnS6 (Fig. 2a and b respectively). The quantification of the three elements provided the following stoichiometry for the pristine and the formamidinium modified crystals: Ag5.88SnS4.42 and Ag4.13SnS3.9 respectively. This implies that the mobile and loosely coordinated silver cations are partially replaced by the organic cation, forming a hybrid material. Argyrodites comprise a large family of compounds with a general chemical formula of A8BX6 (A = Cu, Ag; B = Si, Ge, Sn; and X = S, Se, and Te). The materials, belonging to this family of chalcogenides, are well known as superionic semiconductors.27,28 In their crystal structure, the A+ cations are loosely bound compared to the strong Sn–S covalent bonds. In the work by Heep et al., the bonding environment in Ag8SnS6 was pictured through electron localization function (ELF) and crystal orbital Hamilton population (COHP).29 This revealed that the charge clouds of Sn–S bonds are delocalized, and the Sn is covalently bonded to the S atoms. Furthermore, the Ag atoms are discrete, and no electron was found to be localized near the Ag atoms. In this work, we report a case where the surface is coated either with a carboxylate group and a long amine or the same carboxylate ligand and a short amine. In Fig. 2c the atomic structure diagram of the ATS crystal is presented, alongside the proposed coordination modes in the two aforementioned cases (Fig. 2d).
A more detailed structural characterization was made viable with XPS surface analysis. In Fig. 3a we present the XPS spectra corresponding to the Ag 3d, Sn 3d and S 2p peaks of the Pr–Ag8SnS6 sample.24,30 The Ag 3d spectrum demonstrates two clearly distinguished components related to the Ag 3d5/2 and Ag 3d3/2 core energy levels due to the spin–orbit coupling. Table 1 summarizes and compares the values obtained in our work with those reported in two representative papers on the XPS analysis of Ag8SnS6. The binding energy values of Ag 3d are shifted to lower binding energy values compared to those reported in literature.24,30 This shift is assigned to the increased electron density of the metal cation after functionalization with electron donating amine ligands.31 The analysis reveals that the Sn2+ peak typically located at 485.7 eV is absent, signaling that all tin is in the Sn4+ state. The S 2p spectrum demonstrates binding energies similar to Ag–S bonding similar to Ag2S crystals, while the absence of components belonging to S–O bonds in the range of 165–171 eV excludes the formation of any oxysulfide species.32
![]() | ||
Fig. 3 XPS spectra of the (a) Ag 3d, (b) Sn 3d and (c) S 2p peaks recorded from the sample Pr–Ag8SnS6. |
The XRD pattern of the Pr–Ag8SnS6 nanoparticles reveals the typical diffraction expected from its canfieldite orthorhombic phase. Ag8SnS6 crystallizes in the orthorhombic Pna21 space group with a crystal structure involving eight inequivalent Ag+ sites. There is a spread of Ag–S bond distances ranging from 2.44 to 2.95 Å. After the ligand exchange with FA+ (Fig. 4a), we observed that the patterns of Pr–Ag8SnS6 and FA–Ag8SnS6, are identical, with the only exception of a slight shift towards smaller d-spacings for the [603] diffraction peak (Fig. 4b). After the ligand exchange, the FTIR spectra (Fig. 4c) demonstrates the stronger presence of amine groups, and a reduced intensity of the aliphatic chains, an observation based on the reduced intensity of the –C–H groups, while both the –N–H bending and the –N–H stretching, that relate to primary amines, appear to shift towards higher energies. This is outlined in Table 2. Furthermore, the imine CN mode of the formamidinium cation appears to shift at higher energies after surface functionalization. Specifically, the broad peak at 1683 cm−1 of the pristine ligand, is observed at 1703 cm−1 in the case of the FA–Ag8SnS6 sample. Contact angle measurements with water, demonstrated that after the cation exchange, the particles are becoming significantly less hydrophobic, something that proves the successful substitution of the aliphatic surfactants with the more polar formamidium cations. The images can be seen in Fig. 4d and reveal a decrease of the contact angle from 97.89° for the Pr–Ag8SnS6 to 38.64° for the FA–Ag8SnS6.
Sample | N–H stretching (cm−1) | N–H bending (cm−1) | C–H bending (cm−1) | C–N (cm−1) |
---|---|---|---|---|
Pr–Ag8SnS6 | — | — | 1461 | — |
FA–Ag8SnS6 | 3393/3571 | 1709 | 1458 | 1123 |
FA–acetate | 3204/3420 | 1683 | — | 1079/1012 |
The UV-visible absorption spectra of the nanoparticles can be seen in Fig. 5a. The shoulder located at 710 nm is similar to the one reported in the work of Liu et al.27 Calibration curves were constructed based on the absorbance at λ = 450 nm and can be seen in Fig. 5b. For the construction of the calibration curves, we subtracted the value at λ = 1200 nm from the absorbance intensity values to avoid the contribution of scattering. Nanomaterials exhibit a very high surface to volume ratio and thus the percentage of atoms on the surface is substantial and the presence of defects and the difference in the dielectric environment greatly affect their electronic, optical, thermal and chemical properties.33 For example, halide perovskites functionalized with sodium dodecylsulfonate (SDS), exhibited significantly increased fluorescence intensity and an external quantum yield of 8.4%.34 When used in light emitting diodes, the SDS molecules have stronger absorption energy on CsPbI3 perovskites compared to oleic acid and thus, they suppress the defect formation due to ligand loss during the nanoparticles purification process. Furthermore, plasmonic chalcogenides such as CuFeS2 demonstrated a red shift of the surface plasmon resonance after electrostatic interaction with covalent organic frameworks and sodium dodecylsulfonate.22 Interestingly, after ligand exchange of the bulky and aliphatic oleylamine with the small and electron dense formamidinium cation, we observed an increase on the absorption and extinction coefficient values, see Fig. 5c and d. This is in accordance with a previous work by Kroupa et al. who studied the optical absorption enhancement in PbS NPs upon ligand exchange from oleate to a series of cinnamate capping agents. Through experimental work and ab initio simulations, the authors concluded that the optical absorption enhancement was due to a coupling between the NPs energy levels of the ligand considering the ligand-nanoparticles dyad as a distinct chemical system.35 For clarity, we present the ratio between the absorption coefficient of the two samples (εFA/εPr) in the right axis of Fig. 5c. We observed an absorption coefficient 2.4 times higher in the case of the modified sample close to the band gap (λ = 950 nm). In the same logic with the fabrication of the calibration curves, the absorption values at low energies corresponding to scattering were subtracted throughout the spectral range. The extinction coefficient κ, was calculated via the equation ; ε is the molar absorption coefficient, λ is the wavelength in cm. Based on the FTIR observations that were earlier discussed, we assign this increase to the strong binding of the electron donating formamidinium ligand which, in addition, does not possess long aliphatic chains like the oleylamine. Similar phenomena have been also observed in BaTiO3–xBi2O3 ceramics, where the absorption increased with increasing the percentage of bismuth.36
![]() | ||
Fig. 5 (a) UV-vis spectra and the corresponding calibration curves (b). Absorption coefficient (ε in M−1 cm−1) (c) and extinction coefficient (k in M−1) (d) values in the range 380–1200 nm for the Pr–Ag8SnS6 and FA–Ag8SnS6. The ratio between εFA/εPr can be seen in Fig. 5c. |
The argyrodite Ag8SnS6 is considered an ideal band gap semiconductor for photovoltaic applications. We proceeded to identify the energy levels and the Fermi level position of both the Pr–Ag8SnS6 and the FA–Ag8SnS6 samples through diffuse reflectance spectra and Ultraviolet Photoelectron Spectroscopy. The Tauc plots obtained from the diffuse reflectance values can be seen in Fig. 6a. After the exchange of organic ligands with the formamidinium cations, we observe a slight increase of the band gap from 1.35 eV to 1.4 eV, with a negligible Urbach tail. The lack of the exponential Urbach tail indicates the absence of amorphous, largely disordered areas.37 Finally, the excitation dependent photoluminescence mapping for the Pr–A8SnS6 can be seen Fig. 6aii. No detectable photoluminescence was observed throughout the ultraviolet and the visible light range. The same behavior was observed for the FA–Ag8SnS6. Preliminary measurements up to λ = 1600 nm again did not reveal any photoluminescence at room temperature.
Ultraviolet Photoelectron Spectroscopy is a powerful tool for the calculation of the valence band, the work function and the Fermi level of a material. In order to record the spectrum, the samples were drop-casted on conductive substrates of FTO. In Fig. 6b the UPS spectra of the two samples are presented. The work function (Wf) is calculated from subtracting the secondary energy cutoff from the He I radiation of 21.22 eV.38 In the case of the Pr–Ag8SnS6, the Fermi level is located 0.66 eV higher than the valence band, while the work function was calculated at 4.46 eV. Taking into consideration the band structure of the Pr–Ag8SnS6 nanocrystals, we observe that the Fermi level lies in between the valence and conduction band. Interestingly, the coordination with the smaller and highly basic formamidinium alongside the partial removal of silver cations, is leading to a re-alignment of the energy bands, with the Fermi level now lying 1.02 eV higher than the valence band, transforming the surface of the nanoparticles into a highly n-type semiconductor.
Finally, we proceeded with the fabrication of thin films photovoltaic devices. SnO2 was employed as an electron transport layer on ITO-covered glass substrates. According to the UV-visible reflectance spectra and the UPS analysis, the band structure of the nanoparticles was calculated and in Fig. 7a, the correlation between the energy levels of all components of the device is depicted. The exact structure of the photovoltaic device can be seen in Fig. 7b. The long aliphatic chains in the case of Pr–Ag8SnS6 were removed after exchange with the short ligand 3-mercaptopropionic acid. Spiro-MeOTAD was employed as the hole transport layer. In the previous work by Zhu et al.20 regarding the use of ATS as a photoabsorber in solid-state solar cells, titanium dioxide (TiO2) was used as an electron transport layer and Spiro-OMeTAD as the hole transport layer, the former being deposited on fluorinated tin oxide (FTO) glass. The band alignment between the nanocrystals, the ETL SnO2 and the HTL Spiro-OMeTAD39 clearly demonstrates that the position of the valence band maximum and the conduction band minimum of the Pr–Ag8SnS6, enables the electron and hole transfer and hence a charge separation and transport towards the electrodes can be achieved. This appears to be hindered in the FA modified sample. The J–V curves for both samples can be seen in Fig. 7c, with the corresponding Voc, Jsc, PCE and FF values are gathered in Table 3. We attribute the slightly higher PCE and FF values of the Pr–Ag8SnS6 to its ability to form more uniform films after spin-coating, due to the long aliphatic chains present on its surface. These long insulating chains do not significantly affect the conductance of the material, since they are exchanged with the shorter 3-mercaptopropionic acid after the thin film formation. Furthermore, the UPS studies revealed that the energy levels of the formamidinium modified nanoparticles moved towards more negative values, lying exactly in the level of the conduction band of the electron transport layer, SnO2. However, we argue that the slightly lower current and fill factor, extracted from the FA-based compound, is due to the high n-type character of the surface of the film; in a n–i–p solar cell configuration we have adopted here, a p-type surface would be better for hole extraction towards the HTM. The Pr–Ag8SnS6 photovoltaic performance was tested 2 months after the fabrication and presented an efficiency of 0.17% (Fig. S1, ESI†). Despite a severe FF loss, the initial PCE remains almost unchanged. This is showing promise for the real-life applications of the absorber.
![]() | ||
Fig. 7 (a) The energy levels alignment. The band structure of both nanoparticles is included in the graph (b) schematic representation of solar cell devices (c) J–V curves. |
Sample (solvent) | J SC (mA cm−2) | V OC (V) | PCE (%) | FF (%) |
---|---|---|---|---|
Pr–Ag8SnS6 | 1.35 | 0.36 | 0.22 | 44.5 |
FA–Ag8SnS6 | 1.17 | 0.31 | 0.16 | 43.3 |
Footnote |
† Electronic supplementary information (ESI) available: A comparison of the J–V curves of the same photovoltaic device after one day and after 2 months of fabrication are presented. See DOI: https://doi.org/10.1039/d5tc00397k |
This journal is © The Royal Society of Chemistry 2025 |