Cesium Tin-Germanium Tri-Iodide (CsSnGeI3)

• CsSnGeI3 is a lead-free perovskite alloy offering tunable bandgaps and improved stability via Sn–Ge compositional adjustment.
• It exhibits direct bandgap semiconducting behavior with bandgaps ranging from 1.33 eV to 1.93 eV, underpinning versatile optoelectronic applications.
• Device simulations reveal its potential in plasmon-enhanced NIR PeLEDs and high-efficiency photovoltaics by optimizing alloy ratios.

Cesium tin-germanium tri-iodide, conventionally written as CsSnGeI$_3$, denotes a lead-free all-inorganic halide perovskite composition within the alloy series CsSn$_x$Ge$_{1-x}$I$_3$ or, equivalently in another notation, CsSn$_{1-x}$Ge$_x$I$_3$. Across recent simulation-driven studies, the term is used in two closely related but not identical ways: as a shorthand for the mixed Sn–Ge iodide family, and, in the context of near-infrared perovskite light-emitting diodes, specifically for the equiatomic alloy $\mathrm{CsSn_{0.5}Ge_{0.5}I_3}$ (Debnath et al., 19 Dec 2025). The material class is studied primarily because it replaces toxic Pb with Sn and Ge, retains direct-gap semiconducting behavior, permits composition-dependent bandgap tuning, and is presented as offering improved stability through Ge incorporation and native-oxide passivation in comparison with purely Sn-based analogues (Hasan et al., 11 Feb 2025).

1. Composition, notation, and phase description

The relevant compositional family is the lead-free perovskite alloy series CsSn$_x$Ge$_{1-x}$I$_x$0, with $_x$1 and $_x$2, spanning the endpoints CsGeI$_x$3 and CsSnI$_x$4 (Debnath et al., 19 Dec 2025). In an alternative notation, the same substitutional chemistry is written as CsSn$_x$5Ge$_x$6I$_x$7, emphasizing Ge substitution on the B site of the cubic perovskite lattice (Hasan et al., 11 Feb 2025). The naming “CsSnGeI$_x$8” is therefore potentially ambiguous unless the compositional convention is stated explicitly.

A central clarification is that, in the plasmon-enhanced PeLED study, CsSnGeI$_x$9 is not a separate stoichiometric endpoint but the equiatomic alloy composition $_{1-x}$0 (Debnath et al., 19 Dec 2025). By contrast, the photovoltaic studies use “CsSnGeI$_{1-x}$1” more broadly as the absorber designation for the mixed Sn–Ge system, while also analyzing the full substitutional range (Sami et al., 30 Jul 2025). A plausible implication is that literature comparisons require careful normalization of notation before interpreting reported bandgaps, transport assumptions, or device metrics.

For the DFT-relaxed structures reported for the alloy system, the undoped compound is cubic Pm3m (221) and the doped structures are reported as Fm3m (225) (Hasan et al., 11 Feb 2025). The same work evaluates the tolerance factor criterion in the stated stable cubic range $_{1-x}$2, with reported values of 0.96 for CsSnI$_{1-x}$3, 0.98 for CsSn$_{1-x}$4Ge$_{1-x}$5I$_{1-x}$6, 0.98 for CsSn$_{1-x}$7Ge$_{1-x}$8I$_{1-x}$9, 1.00 for CsSn$_3$0Ge$_3$1I$_3$2, and 0.99 for CsGeI$_3$3, supporting the structural viability of the full series (Hasan et al., 11 Feb 2025).

2. Electronic structure and composition-dependent optical response

The dominant electronic trend across the alloy series is a monotonic increase in bandgap with increasing Ge content. Using HSE06, the reported direct bandgaps are 1.331 eV for CsSnI$_3$4, 1.483 eV for CsSn$_3$5Ge$_3$6I$_3$7, 1.695 eV for CsSn$_3$8Ge$_3$9I$_{1-x}$0, 1.840 eV for CsSn$_{1-x}$1Ge$_{1-x}$2I$_{1-x}$3, and 1.927 eV for CsGeI$_{1-x}$4 (Hasan et al., 11 Feb 2025). The PeLED study reports the same end-member trend, describing the bandgap as increasing from 1.331 eV for CsSnI$_{1-x}$5 to 1.927 eV for CsGeI$_{1-x}$6 with increasing Ge content (Debnath et al., 19 Dec 2025).

All studied compounds are described as direct bandgap semiconductors (Hasan et al., 11 Feb 2025). The reported interpretation is that, as Ge content increases, the conduction band minimum shifts upward while the valence band maximum remains nearly fixed, so the gap widens. The physical rationale given is the smaller ionic or atomic size of Ge relative to Sn, shorter interatomic distances, stronger binding of valence electrons, and a higher excitation energy required for promotion to the conduction band (Hasan et al., 11 Feb 2025).

The density-of-states analysis attributes the VBM mainly to Sn/Ge $_{1-x}$7 orbitals and I $_{1-x}$8 orbitals, while the CBM is mainly derived from B-site $_{1-x}$9 orbitals, namely Sn 5$_x$0 or Ge 4$_x$1, with I contributions near the edge (Hasan et al., 11 Feb 2025). The same work notes that Ge incorporation increases DOS near the band edges, strengthens band-edge localization, and is associated with Urbach-tail-like band-edge broadening (Hasan et al., 11 Feb 2025). This suggests that compositional tuning affects not only the fundamental gap but also near-edge optical broadening and defect sensitivity.

The dielectric response is written as

$_x$2

with $_x$3 describing polarization or energy storage and $_x$4 describing dissipation or absorption (Hasan et al., 11 Feb 2025). In the DFT-informed PeLED analysis, wavelength-dependent refractive index $_x$5 and extinction coefficient $_x$6 are obtained from the dielectric function through

$_x$7

$_x$8

and the absorption coefficient is expressed as

$_x$9

The same study also gives the Kramers–Kronig relation

$_3$0

Across the alloy series, the refractive index is reported to remain high, roughly 2.2–2.6 in the visible/NIR region, which is favorable for strong optical confinement but detrimental to planar outcoupling (Debnath et al., 19 Dec 2025).

The photovoltaic DFT study reports static dielectric constants $_3$1 of 5.47, 3.80, 3.89, 3.98, and 5.95 for CsSnI$_3$2, CsSn$_3$3Ge$_3$4I$_3$5, CsSn$_3$6Ge$_3$7I$_3$8, CsSn$_3$9Ge$\mathrm{CsSn_{0.5}Ge_{0.5}I_3}$0I$\mathrm{CsSn_{0.5}Ge_{0.5}I_3}$1, and CsGeI$\mathrm{CsSn_{0.5}Ge_{0.5}I_3}$2, respectively, with corresponding static refractive indices $\mathrm{CsSn_{0.5}Ge_{0.5}I_3}$3 of 2.50, 1.99, 2.01, 2.03, and 2.63 (Hasan et al., 11 Feb 2025). Reported static reflectivities $\mathrm{CsSn_{0.5}Ge_{0.5}I_3}$4 are 20%, 11%, 11.9%, 12.3%, and 21.8% for the same sequence, and the paper states that Ge substitution improves the absorption profile in the visible spectrum while reducing reflectivity relative to pristine CsSnI$\mathrm{CsSn_{0.5}Ge_{0.5}I_3}$5 over the useful spectral range (Hasan et al., 11 Feb 2025).

3. Near-infrared PeLEDs and plasmon-assisted outcoupling

In the NIR PeLED context, the alloy series is studied as a tunable emitter platform whose appeal is threefold: lead-free composition, NIR emission, and stability tuning through Ge incorporation (Debnath et al., 19 Dec 2025). The study explicitly motivates the work by the high refractive index of these materials, which makes them attractive emitters but poor planar light extractors. The methodology therefore combines DFT for composition-specific optical constants with FDTD for plasmon-assisted device optimization (Debnath et al., 19 Dec 2025).

The simulated PeLED stack is

$\mathrm{CsSn_{0.5}Ge_{0.5}I_3}$6

with radiative recombination represented by a dipole emitter in the perovskite layer (Debnath et al., 19 Dec 2025). The plasmonic element is an Au/SiO$\mathrm{CsSn_{0.5}Ge_{0.5}I_3}$7 core-shell nanorod near the ZnO/perovskite interface. The Au core provides the localized surface plasmon resonance, the SiO$\mathrm{CsSn_{0.5}Ge_{0.5}I_3}$8 shell protects carriers from quenching, and the nanorod length and radius are tuned to match the LSPR to the emitter spectrum (Debnath et al., 19 Dec 2025). The FDTD setup uses periodic boundary conditions laterally, PML boundaries vertically, and a refined mesh near the nanorod-emitter region.

Three device-level metrics are defined explicitly. The Purcell factor is

$\mathrm{CsSn_{0.5}Ge_{0.5}I_3}$9

where $_x$0 is the decay rate with the plasmonic structure and $_x$1 is the bare-emitter decay rate. The light extraction efficiency is given as

$_x$2

that is, the fraction of generated photons escaping the device. The spectral overlap metric is the cosine similarity

$_x$3

These quantities are used to compare compositions and nanorod geometries within a single plasmonic design framework (Debnath et al., 19 Dec 2025).

For the equiatomic composition $_x$4, identified in that study with the notation CsSnGeI$_x$5, the reported values are: emission peak 731 nm, optimized nanorod geometry 70 nm length and 17 nm radius, Purcell factor 5.3$_x$6, LEE 25%, LEE enhancement 33%, and spectral overlap $_x$7 (Debnath et al., 19 Dec 2025). The same paper identifies this composition as the best-balanced overall performer, arguing that it offers the optimal balance of extraction efficiency, radiative enhancement, spectral overlap, and oxidation stability for wearable and flexible optoelectronic applications.

The principal tradeoff is compositional. Sn-rich compositions provide stronger emission-rate enhancement and deeper NIR emission, but they also have higher refractive index and stronger photon trapping. Ge-rich compositions improve extraction and stability but shift emission away from the desired NIR region (Debnath et al., 19 Dec 2025). The paper therefore distinguishes between the composition favored for maximum spontaneous-emission enhancement and the composition favored for overall system-level LED performance.

4. Comparative composition dependence in PeLED design

The composition-specific optimization reveals that distinct alloy ratios maximize different figures of merit. For CsSn$_x$8Ge$_x$9I$_{1-x}$0, the study reports the highest Purcell factor, namely 12.1$_{1-x}$1, with emission about 674 nm, LEE 19%, and spectral overlap 0.80 (Debnath et al., 19 Dec 2025). This composition is recommended when spontaneous emission rate is the primary design objective.

For CsSnI$_{1-x}$2, corresponding to $_{1-x}$3 in the CsSn$_{1-x}$4Ge$_{1-x}$5I$_{1-x}$6 notation, the same study reports the most strongly NIR-shifted emission, about 931 nm, together with Purcell factor 8.0$_{1-x}$7, LEE 17.5%, LEE enhancement 36%, and spectral overlap 0.96 (Debnath et al., 19 Dec 2025). The paper describes Sn-rich compositions as achieving spectral overlap as high as 96%, but also as suffering from poorer overall extraction because of high index and stronger light trapping.

By contrast, the equiatomic composition $_{1-x}$8 gives the highest reported LEE = 25% while preserving strong Purcell enhancement and high spectral overlap (Debnath et al., 19 Dec 2025). The interpretation offered is explicitly a tradeoff rather than a monotonic compositional improvement. This is significant because it rules out a simple “more Sn for better NIR” or “more Ge for better extraction” heuristic. Instead, the design rule is to use DFT-calculated optical constants to match the chosen alloy composition with plasmonic nanorod geometry, then select the composition according to the relevant priority: $_{1-x}$9 for maximum emission-rate enhancement, $_x$00 for the best overall balance, and $_x$01 for the strongest NIR character and highest spectral overlap (Debnath et al., 19 Dec 2025).

A common misconception would be to treat “best composition” as a universal statement independent of application. The published results do not support that simplification. They instead separate best Purcell factor from best balanced device, assigning those roles to CsSn$_x$02Ge$_x$03I$_x$04 and CsSn$_x$05Ge$_x$06I$_x$07, respectively (Debnath et al., 19 Dec 2025).

5. Single-junction photovoltaic modeling and Ge-substitution optimization

In photovoltaic modeling, the same mixed Sn–Ge system is investigated as a lead-free absorber whose electronic structure and optical spectra are tuned by B-site substitution and then transferred into device simulation (Hasan et al., 11 Feb 2025). The DFT workflow uses VASP, GGA-PBE, HSE06, PAW potentials, a 500 eV plane-wave cutoff, and $_x$08-meshes of 6×6×6 for the pure structure and 3×3×3 for doped supercells. The authors build a cubic CsSnI$_x$09 cell, expand it to a 2×2×2 supercell, and replace Sn with Ge at 25%, 50%, 75%, and 100% (Hasan et al., 11 Feb 2025). They then import DFT-derived absorption spectra and bandgaps into SCAPS-1D, explicitly because SCAPS’s built-in optical absorption is regarded as too simplified.

The simulated n–i–p device stack is FTO / PCBM / CsSn$_x$10Ge$_x$11I$_x$12 / Cu$_x$13O / Au, with thicknesses of 200 nm for FTO, 50 nm for PCBM, 500 nm for the absorber, 100 nm for Cu$_x$14O, and 60 nm for Au (Hasan et al., 11 Feb 2025). The work functions are 4.4 eV for FTO and 5.45 eV for Au. The model includes a Gaussian defect distribution in the absorber, interface defects at both transport-layer interfaces, interface defect levels centered around 0.6 eV above the VB, a series resistance of 4200 $_x$15, and a shunt resistance of 1 $_x$16.

The principal device-level result is that performance improves with Ge substitution up to an optimum at 75% Ge concentration, corresponding to CsSn$_x$17Ge$_x$18I$_x$19 (Hasan et al., 11 Feb 2025). The reported optimized metrics are PCE = 23.8%, $_x$20 V, $_x$21 mA/cm$_x$22, and FF = 87.80%. The same work states that this raises the standalone Sn-based cubic iodide perovskite efficiency from 10.5% to 23.8% (Hasan et al., 11 Feb 2025).

The improvement is explained as a joint consequence of bandgap tuning, improved band alignment, enhanced visible absorption, reduced reflectivity, and better carrier generation (Hasan et al., 11 Feb 2025). The authors also state that 25%, 50%, and 75% Ge doping have the strongest and nearly identical absorption in the visible range, leading to high $_x$23 and high PCE, whereas the 0% and 100% Ge endpoints are less optimal because their visible absorption is poorer. This is a second instance, distinct from the PeLED results, in which the optimal composition is a mixed alloy rather than an end member.

The same study complements SCAPS with FDTD simulations in Ansys Lumerical 2022, examining electric-field distributions at 300 nm, 550 nm, 705 nm, 1320 nm, and 2000 nm (Hasan et al., 11 Feb 2025). It reports that field intensity and absorption are especially strong for the 50–75% Ge-doped absorbers, that the generation rate is highest at the ETL/perovskite interface and especially strong for 50% Ge doping near 630 nm, and that the device temperature rises by about 15.52 K from ambient 300 K due mainly to SRH recombination and thermalization losses (Hasan et al., 11 Feb 2025). These results suggest that the compositionally optimized absorber also affects internal field localization, generation nonuniformity, and thermal load.

6. Tandem perovskite/silicon architectures and system-level significance

A separate tandem-solar-cell study uses CsSnGeI$_x$24 as the top-cell absorber in a fully inorganic, lead-free perovskite/silicon heterojunction tandem (Sami et al., 30 Jul 2025). The material is selected because it addresses lead toxicity, is fully inorganic, has a suitable bandgap for tandem operation with silicon, and is described as having good carrier transport and absorption. The same paper notes prior literature indicating native-oxide passivation and improved stability for related CsSnGeI systems (Sami et al., 30 Jul 2025).

The tandem architecture is a monolithic two-terminal stack in which CsSnGeI$_x$25 absorbs the high-energy part of the spectrum while the silicon bottom cell absorbs longer-wavelength photons. The top-cell configuration is given explicitly as FTO / TiO$_x$26 / CsSnGeI$_x$27 / Cu$_x$28O, connected to the bottom subcell by an ITO recombination layer (Sami et al., 30 Jul 2025). The full front-to-back stack is: Al finger electrode, SiO$_x$29 ARC, FTO, TiO$_x$30, CsSnGeI$_x$31, Cu$_x$32O, ITO, n-doped a-Si, GaSb auxiliary absorber, c-Si, p-doped a-Si, SiO$_x$33 rear passivation, Si$_x$34N$_x$35 containing cylindrical Au nanorods, and an Ag mirror / back contact (Sami et al., 30 Jul 2025).

The device is analyzed with a coupled FDTD + drift-diffusion workflow. For the optical simulation, Maxwell curl equations are solved under AM 1.5G, with PML boundaries on top and bottom and periodic boundaries laterally. The absorbed power density is written as

$_x$36

the local carrier-generation rate as

$_x$37

and the total generation as

$_x$38

The electrical model then solves the Poisson equation, current equations, and continuity equations self-consistently to obtain $_x$39-$_x$40 curves and derived figures of merit (Sami et al., 30 Jul 2025).

The optimized CsSnGeI$_x$41 absorber thickness is 160 nm, with an optimal p-doping concentration of $_x$42 (Sami et al., 30 Jul 2025). The paper further reports that PCE is mainly determined by CsSnGeI$_x$43 thickness, that $_x$44 peaks around 150–200 nm, and that $_x$45 and fill factor decrease as the perovskite becomes too thick. For transport layers, TiO$_x$46 is selected because of a favorable conduction-band offset and Cu$_x$47O because it has the most favorable valence-band offset among the compared HTLs, with reported negative VBOs of −0.03 eV for Cu$_x$48O, −0.14 eV for PTAA, −0.14 eV for NiO, and −0.32 eV for CuI, and negative CBOs of −0.36 eV for ZnO, −0.19 eV for ZnSe, and −0.10 eV for both TiO$_x$49 and PCBM (Sami et al., 30 Jul 2025).

Under maximum power point tracking, the top cell alone achieves PCE 23.46%, $_x$50 V, $_x$51 mA/cm$_x$52, and FF 87.18%. The optimized tandem device achieves PCE 34.93%, $_x$53 V, $_x$54 mA/cm$_x$55, and FF 84.74% (Sami et al., 30 Jul 2025). The tandem current equals the top-cell current, consistent with the series-connected relation

$_x$56

The paper therefore identifies CsSnGeI$_x$57 as the current-limiting junction in the optimized stack.

The GaSb auxiliary absorber enhances near-infrared absorption and allows the required c-Si thickness to be reduced to 2 $_x$58m, while the rear plasmonic reflector with cylindrical Au nanorods in Si$_x$59N$_x$60 slightly increases absorption and reduces reflection, especially at longer wavelengths (Sami et al., 30 Jul 2025). However, the same study is explicit that high efficiency is retained even without the plasmonic structure: 34.32% PCE, 1.9334 V, 21.18 mA/cm$_x$61, and 83.79% FF without nanorods, versus 34.93%, 1.9347 V, 21.30 mA/cm$_x$62, and 84.74% FF with them. This yields only a 0.61% absolute PCE drop, so the plasmonic back reflector is described as helpful but not mandatory (Sami et al., 30 Jul 2025).

Taken together, the available studies position CsSnGeI$_x$63 and the broader CsSn–Ge iodide alloy family as a compositionally tunable, lead-free perovskite platform that supports two distinct but related device paradigms: plasmon-enhanced NIR PeLEDs and high-efficiency single-junction or tandem photovoltaics. The recurrent theme is not a single universally optimal stoichiometry, but a composition-dependent compromise among bandgap, absorption, refractive index, outcoupling, band alignment, and oxidation stability (Debnath et al., 19 Dec 2025).