Highly transparent and color-neutral Eu³⁺-doped glass luminescent solar concentrators for scalable BIPV integration

Abstract

As urban design increasingly prioritizes energy efficiency and sustainability, building-integrated photovoltaics (BIPVs) have emerged as a compelling approach for on-site solar energy harvesting. Among these, luminescent solar concentrators (LSCs) offer transparent and architecturally compatible solutions by guiding spectrally shifted light to edge-mounted photovoltaic cells. In this study, we report the first demonstration of rare-earth doped (Eu³⁺) fully inorganic glass based LSCs, combining high optical performance, long-term environmental stability, and process scalability. By tuning the Eu³⁺ concentration, waveguiding losses are minimized, and the emission efficiency and photon transport are maximized. Devices fabricated with the optimal glass composition (2.5 mm thick; 2 to 6 × 6 cm²) exhibit outstanding average visible transmittance (AVT ≈ 90%) and near-neutral color rendering (CRI ≈ 98), enabling seamless integration into modern architectural environments. The glass matrix maintains its optical and structural integrity under thermal, chemical, and mechanical stress, confirming its long-term durability for real-world applications. External photon efficiency remains stable at ∼6.4% across all device sizes, while the highest power conversion efficiency (PCE) of 0.852% in the 4-edge configuration is achieved for the most compact device. These findings position Eu³⁺-doped glass as a robust, scalable, and multifunctional platform for the next generation of BIPV-integrated LSCs, offering a rare-earth-based solution for durable, visually neutral solar harvesting surfaces.

---

1. Introduction

As urban populations grow and transportation systems electrify, cities are becoming focal points of global energy demand.¹ This transformation highlights the importance of energy strategies that are both sustainable and spatially integrated into the built environment.² In this context, on-site energy generation and consumption, where energy is produced and used within the built environment, is increasingly viewed as a promising strategy for improving urban energy resilience.³ Herein, building-integrated photovoltaics (BIPVs) offer a promising solution by enabling architectural surfaces such as windows and façades to function as energy-harvesting elements. Nevertheless, the requirement for light transmission and visual clarity on these surfaces presents a challenge for integrating conventional photovoltaic modules, which are inherently opaque.⁴ To address this, a wide variety of transparent and semi-transparent photovoltaic technologies have been developed, including those based on thin-films, wavelength-selective organic and perovskite materials, and even micro-structured crystalline silicon capable of achieving high efficiency with neutral coloration. A common architectural feature of these approaches, however, is that the active, power-generating layers and associated electrodes are distributed across the entire viewing area. This co-location creates a fundamental trade-off between power conversion efficiency (PCE), visible transmittance, and aesthetic quality, often resulting in compromised color neutrality, visible grid lines, or an inhomogeneous appearance that can hinder seamless architectural integration. In this context, luminescent solar concentrators (LSCs) represent a distinct architectural paradigm that avoids this trade-off by collecting light through a large transparent waveguide and converting it to electricity using small photovoltaic cells positioned at the edges.^5,6 This unique architecture enables uniform transparency and customizable aesthetics without visually disruptive components in the line of sight, making LSCs a particularly compelling solution for applications where pristine visual quality and design flexibility are paramount.⁷

LSCs operate by embedding luminescent species within an optical waveguide, where incident light is collected through the top surface, spectrally shifted, and redirected toward PV cells mounted along the lateral edges via total internal reflection. Since the PV cells are confined to the edges, the device's overall transparency and color perception are primarily governed by the spectral characteristics of the luminescent species.^8–10 For optimal energy harvesting, broadband absorbers are often employed;¹¹ however, these materials typically absorb part of the visible spectrum, resulting in a trade-off between power output and optical transparency.¹² To overcome this limitation, trivalent europium (Eu³⁺) emerges as a promising dopant due to its sharp, narrow-band absorption features. These enable high visible light transmittance while still supporting efficient photoluminescence (PL), making Eu³⁺-based systems particularly suitable for visually neutral and transparent LSC applications.¹³

Over the past years, various LSC systems have been explored, including planar,¹⁴ cylindrical,¹⁵ and fiber-shaped^16–18 geometries. These systems employ a broad range of europium-based luminescent species such as organic dyes,¹⁹ coordination complexes,^20–27 and rare-earth salts,^28–31 typically incorporated via surface coatings on transparent substrates or dispersed within polymer-based waveguides. While these approaches have demonstrated the photophysical potential of Eu³⁺-containing luminophores, they suffer from key limitations. Most notably, polymeric hosts offer limited surface finish versatility and are inherently vulnerable to environmental degradation. The use of polymer-based materials in building facades raises concerns regarding long-term weathering, fire resistance, and mechanical durability.³² Surface-coated devices are susceptible to delamination and optical loss over time, reducing their viability for outdoor deployment.³³ Furthermore, many of these systems require complex, multi-step fabrication processes—such as ligand engineering or matrix functionalization—hindering scalability and widespread adoption.³³

These issues are especially critical for BIPV applications, where materials must meet not only optical and environmental demands but also stringent structural and safety standards. In contrast, Eu³⁺-doped glasses offer superior thermal, mechanical, and chemical stability, ensuring long-term reliability.³⁴ Importantly, Eu³⁺ ions can be directly incorporated into conventional glass production workflows, allowing seamless integration with large-scale flat-glass manufacturing.³⁵ Given that glass is already a preferred architectural material due to its durability, fire resistance, and aesthetic versatility, Eu³⁺-doped glasses represent a promising, underexplored option for LSCs. Despite their promise, the development of fully inorganic, transparent LSCs that combine high optical performance with long-term environmental durability remains unrealized. In particular, the integration of rare-earth luminophores directly into stable glass matrices has not yet been explored, leaving a critical gap in the advancement of robust and scalable BIPV-compatible devices. Addressing this challenge, the present study investigates Eu³⁺-doped glasses as an alternative material platform for LSCs, leveraging the inherent advantages of glass—including its chemical resilience, structural integrity, and compatibility with large-scale fabrication. By systematically examining the photoluminescence and waveguiding behavior of Eu³⁺-doped glasses across a range of dopant concentrations, this work aims to establish the viability of a new class of durable, color-neutral LSCs suitable for architectural integration. In doing so, it lays the groundwork for overcoming key limitations of polymer-based systems and advancing next-generation luminescent materials for energy-harvesting surfaces.

2. Experimental

2.1. Glass synthesis

Selecting an appropriate glass matrix and optimal Eu³⁺ concentration is crucial for balancing luminescence and waveguiding properties, thereby determining the practical applicability of such glasses in LSCs. Drawing from our previous experience, a tailored nominal glass composition is selected as 5CaO–3PbO–6Na₂O–12ZnO–2Al₂O₃–32B₂O₃–40SiO₂–(xEu₂O₃, x = 0, 0.25, 0.5, and 1) (mol%) (corresponding to approximately 1.06, 2.13, and 4.26 wt% of Eu³⁺, respectively), which allows for systematic investigation of the effects of Eu³⁺ concentration on both optical performance and waveguiding behavior. High-purity raw materials of analytical reagent grade (>99.9%), including CaCO₃, PbO, Na₂CO₃, ZnO, Al₂O₃, H₃BO₃, SiO₂ and Eu₂O₃, are procured from Sigma-Aldrich and Alfa Aesar and used without further purification. The materials are precisely weighed (±0.0001 g), homogenized in an agate mortar and pestle, and transferred to a platinum crucible. The batch mixture is melted at 1250 °C for 2 hours under ambient conditions. The molten glass is initially cast onto a preheated stainless-steel mold. The resulting glass is then crushed, ground into a fine powder, and remelted. This melting and grinding cycle is repeated several times to ensure thorough homogenization. Finally, the glass is annealed at 450 °C for 5 hours to relieve internal stress, followed by a controlled cooling process to room temperature. The glass samples are then cut into square prisms with 2 cm sides, and all surfaces are ground and polished to ensure optical smoothness and a uniform thickness of 2.5 mm. To systematically evaluate the effect of the size on LSC performance, glass samples doped with a fixed concentration of Eu₂O₃ (0.5 mol%) are prepared using the same procedure, but with increased surface areas of 4 × 4, and 6 × 6 cm².

2.2. Characterization studies

The thermal properties of the as-cast glass samples are analyzed using a Netzsch STA 449 F3 Jupiter instrument. Approximately 25 ± 1 mg of powdered sample is sealed in aluminum pans and heated at a rate of 10 °C min⁻¹ under a constant argon flow to determine T_g. XRD analysis is performed using a Malvern PANalytical Empyrean multicore high-performance diffractometer with Cu K_α1 and Cu K_α2 radiation (λ = 0.15418 nm, 45 kV, 40 mA). Data are collected over a 2θ range of 10° to 90° with a step size of 0.02° to assess the amorphous structure of the glass samples. Optical transmittance and absorption spectra are recorded using an Edinburgh Instruments DS5 UV-vis spectrophotometer.

Steady-state PL spectra of the glass samples are recorded using an Edinburgh Instruments FS5 fluorescence spectrofluorometer, equipped with a 150 W xenon lamp (signal-to-noise ratio >10 000 : 1 for the water Raman signal). Measurements are performed using a PMT-1010 detector, with a scan slit of 1 nm, a fixed offset slit of 1 nm, and no optical filters. Distance-dependent edge emission measurements are recorded using a CNI Arora 4000 high resolution spectrophotometer. The instrument is equipped with a fiber optic cable from the same manufacturer, featuring a core diameter of 200 µm and an entrance aperture of 10 µm. Monochromatic light from an Edinburgh Instruments FS5 spectrofluorometer is used as the excitation source. To improve the signal-to-noise ratio, each spectrum is averaged from five consecutive scans. Photoluminescence quantum yield (n_pl) values are measured under 393 ± 1 nm excitation using an integrating sphere (150 mm internal diameter, polytetrafluoroethylene-coated) with an estimated measurement uncertainty of ±5%. Bulk, surface, and edge emission spectra are collected using the same integrating sphere. A diffuse 393 nm excitation source is used to ensure homogeneous illumination and to minimize directional artifacts. For the bulk emission measurement, no faces of the sample are masked. To isolate the surface emission, the lateral faces are covered with black tape. Finally, the edge emission is calculated by subtracting the surface emission spectrum from the bulk emission. The time-resolved PL decay spectra are acquired using the time-correlated single-photon counting (TCSPC) technique, employing microsecond pulsed excitation (393 ± 1 nm) as the excitation source. The decay curves are fitted to a bi-exponential model to ensure a reduced χ² value close to unity:

where I(t) represents the emission intensity, A₁ and A₁ are decay constants and τ₁ and τ₂ are decay components, respectively. Average lifetime (τ_ave) values are calculated using the following equation:

Temperature-dependent and heating–cooling cyclic PL measurements up to 200 °C are conducted using a Pike Technologies heated solid transmission attachment integrated into the FS5 spectrofluorometer. Photostability is assessed by continuously exposing the selected sample to 393 nm excitation within the FS5 chamber for 50 hours. PL measurements are recorded hourly for the first 10 hours, and subsequently every 10 hours until the end of the test.

Surface hardness is evaluated using the Vickers indentation method with a DHV-1000(Z) tester, averaging five non-overlapping indentations per sample. Applied loads range from 100 to 1000 gf, with a dwell time of 30 seconds. Chemical durability is evaluated by immersing a selected glass sample in deionized water at room temperature for 30 days, with transmittance and n_pl measurements taken every 3 days to monitor moisture-related degradation. The refractive index (n) of each glass sample is measured at 520 nm using an Abbe refractometer. AVT and CRI values are calculated from the measured transmittance spectra in accordance with CIE 13.3-1995 and CIE 15:2018 standards, using spectral data acquired under ASTM E903 measurement conditions.

A monofacial PERC solar cell is laser-cut into strips measuring 8 cm in length and 1 cm in height. I–V measurements for solar cells and LSCs are performed using a Keithley 2410 source meter and an A class ABA Newport solar simulator (1-sun). The extracted photovoltaic parameters for the cut-cell are as follows: an open-circuit voltage (V_oc) of 667 mV, a short-circuit current density (J_sc) of 37.1 mA cm⁻², and a fill factor (FF) of 67.5%. Notably, the PCE decreases from ∼22% (before cutting) to 16.7% post-cutting, due to reduced active areas, edge-related defects, or potential damage to the electrical contacts during the cutting process. All PCE values represent the average of five independent measurements, with a standard deviation of approximately 0.05%.

3. Results and discussion

3.1. Thermal, structural, optical and photoluminescence properties

In this study, Eu³⁺-doped and undoped glass samples are prepared via conventional melt-quenching, as detailed in the Experimental section. The thermal, structural, and optical characteristics of these samples are systematically investigated to assess their suitability as luminescent media and waveguides in LSC applications.

The thermal behavior of the glass samples is investigated using simultaneous thermal analysis (STA) to evaluate their thermal stability and to determine appropriate annealing conditions. The selected 0.50 mol% Eu³⁺-doped sample exhibits a glass transition onset (T_g) at 523 °C, as shown in Fig. 1a. Based on this, annealing is performed at 450 °C for 5 hours—safely below T_g to prevent crystallization. STA results confirm that the glasses exhibit high thermal stability, retaining structural integrity at temperatures relevant to BIPV applications. The amorphous nature of the samples is verified via X-ray diffraction (XRD). The 0.50 mol% Eu³⁺-doped glass sample exhibits a broad amorphous halo, associated with short-range atomic order in the glass network, and no sharp Bragg diffraction peaks (Fig. 1b), confirming the absence of crystalline phases.

Fig. 1 (a) STA curve, and (b) XRD pattern of the selected 0.50 mol% Eu³⁺-doped glass sample. (c) Optical absorption spectra of all glass samples, with the inset highlighting Eu³⁺ absorption features as well as photographs of the samples under daylight and UV light illumination. (d) Steady-state PL excitation and emission spectra of Eu³⁺-doped samples monitored at the 613 nm emission, along with the emission spectrum recorded under 393 nm excitation. (e) η_pl of glass samples doped with 0.25, 0.50, and 1.00 mol% Eu₂O₃ under 393 nm excitation. (f) Time-resolved PL decay curves under 393 nm pulsed excitation, monitoring the 613 nm emission peak.

Fig. 1c displays the optical absorption spectra of all glass samples, with the inset highlighting the weak Eu³⁺ absorption features that are otherwise obscured in the full spectral range. On the right, digital photographs of undoped and Eu³⁺-doped glass samples under both daylight and UV illumination are presented. Under daylight, all samples are optically clear, colorless, and free of visible defects such as bubbles, surface scratches, or inclusions—demonstrating high optical quality. Upon UV excitation, the doped glasses exhibit a uniform and intense red emission across the entire surface—except for the 0.25 mol% Eu³⁺-doped sample, which shows significantly lower intensity. No observable luminescence inhomogeneity or scattering is detected, confirming the optical uniformity of the doped glass matrix.

Additionally, a gradual red-shift of the UV absorption edge is observed with increasing Eu³⁺ concentration, shifting from approximately 338 nm for the undoped glass to 340, 342, and 345 nm for the 0.25, 0.50, and 1.00 mol% Eu³⁺-doped glass samples, respectively. This apparent displacement originates from the increasing cumulative absorbance of the Eu³⁺ 4f–4f transitions, which slightly elevates the baseline in the near-UV region and effectively shifts the apparent absorption edge toward longer wavelengths.³⁶ The absorption spectra show seven distinct peaks that appear at 360, 374, 381, 393, 414, 464, and 532 nm, corresponding to transitions from the Eu³⁺ ground state ⁷F₀ to the excited states ⁵L₇, ⁵H₄/⁵H₅, ⁵G₂, ⁵L₆, ⁵D₃, ⁵D₂, and ⁵D₁, respectively. An additional transition at 525 nm, attributed to the ⁷F₀/⁷F₁ to ⁵D₁ transition, is also evident. As expected, overall absorbance increases with Eu₂O₃ concentration, reaching a maximum in the 1.00 mol% doped sample.

To clarify the excitation processes more explicitly, photoluminescence excitation (PLE) measurements are performed enabling the identification of Eu³⁺ excitation transitions that are not clearly visible in the absorption spectrum. Fig. 1d presents the steady-state PLE spectra of the Eu³⁺-doped glass samples monitored at the 613 nm emission, together with the corresponding emission spectrum recorded under 393 nm excitation. The PLE spectra display the characteristic 4f–4f electronic transitions of the Eu³⁺ ions, in addition to a broad ligand-to-metal charge-transfer (CT) band located around 250–300 nm. The distinct 4f–4f transition peaks centered at approximately 318, 361, 381, 393, 414, and 464 nm correspond to the ⁷F₀ → ⁵H_5,6,7, ⁷F₀ → ⁵D₄, ⁷F₀ → ⁵G₂, ⁷F₀ → ⁵L₆, ⁷F₀ → ⁵D₃, and ⁷F₀ → ⁵D₂ transitions, respectively.

The optical absorption and excitation behavior of the Eu³⁺-doped glasses reveal their strong potential for BIPV-oriented LSC applications. As shown in Fig. 1c, the glass matrix exhibits a deep-UV absorption edge (shaded region in the graph), providing intrinsic UV-blocking while maintaining high transparency in the visible region. The weak and narrow 4f–4f absorption bands of Eu³⁺ ensure color neutrality under daylight—an essential advantage for building integration. In combination, the host and Eu³⁺ ions enable partial conversion of UV photons into visible red emission, thereby enhancing solar energy harvesting without compromising optical transparency.

The PL spectra exhibit five characteristic emission bands centered at approximately 578, 592, 613, 653, and 703 nm, which are assigned to the ⁵D₀ → ⁷F_J (J = 0, 1, 2, 3, 4) transitions of Eu³⁺ ions, respectively (Fig. 1d). The high intensity of the peak at 613 nm, corresponding to the normally forbidden ⁵D₀ → ⁷F₂ electric dipole transition, indicates that the Eu³⁺ ions occupy low-symmetry sites within the glass network.³⁷ As the Eu₂O₃ concentration increases from 0.25 mol% to 1.00 mol%, the overall emission intensity is observed to increase. This trend is primarily attributed to the larger population of emitting Eu³⁺ ions incorporated into the glass matrix.

To provide a direct measure of the luminescence efficiency, the absolute photoluminescence quantum yield (η_pl) is determined for each sample. The η_pl values are found to be 40.95%, 70.71%, and 66.94% for the 0.25, 0.50, and 1.00 mol% samples (Fig. 1e), respectively. The quantum yield reaches a maximum at 0.50 mol% before decreasing at the highest concentration, indicating the possible onset of concentration quenching.

Time-resolved decay curves for the ⁵D₀ → ⁷F₂ transition are recorded for the 613 nm emission peak under pulsed 393 nm excitation. Fig. 1f exhibits the average lifetime (τ_ave) as well as the other lifetime parameters including τ₁, τ₂, A₁, A₂, and χ² extracted from the decay profiles, while the fitting procedures and calculation methods are described in detail in the experimental section. τ_ave decreases continuously from 2.36 ms (for 0.25 mol%) to 2.20 ms (for 1.00 mol%), whereas η_pl increases sharply from 0.25 mol% to 0.50 mol%, before slightly decreasing at 1.00 mol%. Based on the standard expressions η_pl = k_R/(k_R + k_NR) and τ_pl = 1/(k_R + k_NR), where k_R and k_NR are the radiative and non-radiative decay rates, respectively, using k_R = η_pl/τ_ave and k_NR = (1 − η_pl)/τ_ave, the PL behavior can be analyzed in terms of competing radiative and non-radiative processes. From 0.25 to 0.50 mol% doping, k_R nearly doubles from 173 s⁻¹ to 309 s⁻¹, while k_NR decreases significantly from 250 s⁻¹ to 128 s⁻¹. This trend suggests that the increase in η_pl is associated with an enhancement of the radiative channel rather than a suppression of non-radiative losses. On the other hand, from 0.50 mol% to 1.00 mol% k_R it decreases slightly from 309 s⁻¹ to 303 s⁻¹, while k_NR increases slightly from 128 s⁻¹ to 150 s⁻¹. At 1.00 mol% doping, non-radiative pathways begin to play a slightly more dominant role, correlating with the moderate reduction observed in η_pl.

3.2. Photonic collection, conversion and waveguiding properties

The optical waveguiding behavior of Eu³⁺-doped glass samples is investigated using PL analysis in conjunction with basic waveguide theory, with the aim of identifying the optimal dopant concentration. Spectral emissions are systematically evaluated under controlled optical conditions from bulk, surface, and edge configurations to assess photon propagation and emission pathways within the glass matrix. Fig. 2a schematically summarizes the main optical events occurring within the Eu³⁺-doped glass waveguides, including: (1) incident light interaction, (2) partial reflection at the air–glass interface, (3) transmission through the glass, (4) absorption by Eu³⁺ luminescent centers, (5) attenuation within the glass host due to minor absorption and scattering, (6) PL emission, (7) escape through the critical angle cone, (8) re-absorption, and (9) total internal reflection. These processes collectively determine the fraction of photons that can be successfully guided to and emitted from the edge surfaces.

Fig. 2 (a) Schematic of the Eu³⁺-doped glass waveguide, illustrating key optical processes including absorption, emission, and photon trapping. (b) PL spectra obtained from bulk, surface, and edge emissions measured in an integrating sphere. The inset tables summarize η_edge, η_int, and η_ext values, with η_ext calculated under both 393 nm excitation and the AM1.5G solar spectrum. On the left, schematic illustrations depict the edge PL measurement configurations for bulk, surface, and edge emissions, with black-taped areas indicating the masked regions. (c) Photographs of 0.50 mol% Eu³⁺-doped glass waveguides with lateral lengths L = 2 cm, L = 4 cm, and L = 6 cm under ambient and UV light. (d) Experimental setup for edge-emission measurements as a function of optical distance (L). (e) Distance-dependent edge PL spectra for the L = 6 cm sample. (f) Exponential fit of integrated edge-emitted PL intensity (I_L) as a function of optical distance for the 0.50 mol% sample.

Fig. 2b presents the PL emission spectra collected from the bulk, surface, and edge of Eu³⁺-doped glass samples, along with the corresponding measurement configurations used to isolate each portion of the emission. In the schematics shown to the left of the spectra, red areas indicate unmasked collection surfaces, whereas black regions represent masked faces that prevent photon output. These configurations are further used to determine the edge photon efficiency (η_edge), enabling a quantitative comparison of the fraction of emitted photon that are successfully guided through the glass matrix and reach the edges. This is defined as:

where I_bulk(λ) and I_edge(λ) represent the PL emission intensities collected from the bulk and edge, respectively. The calculated η_edge values are 77.60%, 77.49%, and 77.30% for the 0.25, 0.50, and 1.00 mol% Eu³⁺-doped glass samples, respectively, indicating high waveguiding efficiency across all concentrations. To better contextualize this metric, one can consider the trajectory of a photon randomly emitted from an excited Eu³⁺ ion—illustrated as step ⑥ in Fig. 2a. If the emission angle is smaller than the critical angle (θ_c ≈ 38.7° for n = 1.60), the photon undergoes total internal reflection (step ⑨) and propagates within the waveguide toward the edges. In an ideal, lossless system, approximately 78.0% of emitted photons would be confined and reach the edges—consistent with Fresnel-based predictions using cos(θ_c).³⁸ In practice, the observed η_edge is influenced by three primary loss mechanisms: (i) reabsorption by other Eu³⁺ ions (step ⑧), (ii) attenuation by the glass host matrix (step ⑤), and (iii) photon escape through the escape cone defined by θ_c (step ⑦). In the present system, reabsorption losses are negligible due to the large Stokes shift (∼1.13 eV), derived from the spectral gap between the most intense excitation (⁷F₀ → ⁵L₆) and emission (⁵D₀ → ⁷F₂) transitions of Eu³⁺, which minimizes spectral overlap. Therefore, the slight decline in η_edge at higher dopant concentrations is primarily attributed to slightly increased host attenuation and/or photon scattering, which enhances vertical photon escape through the top and bottom surfaces—ultimately reducing the fraction of photons guided to the edges. To simultaneously capture both radiative efficiency and waveguiding performance, the internal photon efficiency (η_int) is calculated as:

η_int = η_edgeη_pl (4)

Accordingly, the calculated η_int values are 31.78%, 54.79%, and 51.74% for the 0.25, 0.50, and 1.00 mol% Eu³⁺-doped glass samples, respectively. The highest η_int, observed at 0.50 mol%, reflects an optimal trade-off between radiative efficiency and waveguiding performance, highlighting this concentration as the most effective for integrated photonic applications.

To account for the effects of initial photon interactions, the external photon efficiency (η_ext), defined as the fraction of incident photons emitted through the edge surface, is calculated as:

η^393 nm_ext = η^393 nm_absη_int (5)

where η^393 nm_abs represents the absorptance of the device at the excitation wavelength. Notably, the absorptance is calculated by assuming normal light incidence and specular reflection. These assumptions are justified by the high optical quality and smoothness of the glass samples, where deviations from the actual values are expected to be minimal. Under these conditions, absorptance can be expressed as follows:

η_abs = (1 − R)(1 − e^{−α_393nm×d}) (6)

Here, R = (1 − n)²/(1 + n)² denotes the Fresnel reflection coefficient at the air/glass interface, α_393 nm is the absorption coefficient at the excitation wavelength, and d is the sample thickness. Therefore, η^393 nm_abs is estimated to be 29.61%, 36.34%, and 43.17% for the 0.25, 0.50, and 1.00 mol% Eu³⁺-doped glass samples under 393 nm excitation, respectively. Based on these values, the corresponding η^393 nm_abs values are determined to be 11.18%, 19.89%, and 22.30%, respectively—demonstrating a clear enhancement in edge-emitted photon output with increasing dopant concentration.

While the above analysis defines the total photon flux guided to the edges as the incident flux reduced by cumulative optical losses, a practical LSC operates under the full AM1.5G solar spectrum rather than a single excitation wavelength. Therefore, eqn (6) must be refined based on the solar-weighted absorptance framework to provide a more realistic measure of solar-driven performance. The solar-weighted absorptance approach is a widely adopted method in LSC research and provides highly reliable results for broadband luminophores such as quantum dots,^39–41 where the absorption spectrum of the luminescent centers can be clearly distinguished from that of the host matrix. However, its direct application to Eu³⁺-doped glass systems presents certain limitations. The glass matrix exhibits strong intrinsic UV absorption, where no Eu³⁺-related emission occurs. Furthermore, the absorption contribution of Eu³⁺ cannot be separated by subtracting an undoped reference spectrum, as the doped and undoped matrices are structurally and optically distinct.

Moreover, it is a well-established principle that photon emission occurs from the lowest excited state of a given multiplicity. Consequently, for Eu³⁺ each 4f–4f excitation populates a distinct intermediate energy level that relaxes toward the emissive ⁵D₀ state through non-radiative pathways. The varying efficiencies of these relaxation routes from different initial states are responsible for the different photoluminescence quantum yields observed for each transition. Consequently, to capture this excitation-dependent behavior accurately, the photoluminescence efficiency η_pl is treated as a wavelength-dependent function,²³ η_pl(λ), and set to zero in the non-emissive UV range dominated by glass host absorption. This treatment accounts simultaneously for both the selective excitation dynamics of Eu³⁺ ions and the parasitic absorption of the host, ensuring a physically consistent evaluation of solar-weighted photon efficiency. Then, the refined equation follows:

where the η_abs(λ) term follows the same formulation as previously described, except that the fixed absorption coefficient α_393 nm is replaced by the wavelength-dependent absorption coefficient α(λ). The refractive index of the glass is assumed to remain approximately constant at 1.60 throughout the relevant spectral range. Likewise, η_edge is treated as wavelength-independent since all emitted photons regardless of excitation wavelength experience identical waveguiding conditions. Based on this refined model, η_ext(λ) values are 3.491%, 6.483%, and 6.351% for the 0.25, 0.50, and 1.00 mol% Eu³⁺-doped glass samples, respectively. The 0.50 mol% sample consistently exhibited the highest efficiency values confirming that this composition represents the most optimal concentration for solar-driven photon harvesting and edge-directed emissions.

To evaluate the effect of waveguide size on photon propagation, 0.50 mol% Eu³⁺-doped samples with edge lengths of 4 cm and 6 cm (thickness: 0.25 cm) are also prepared. Fig. 2c exhibits the digital images of these samples taken under both daylight and UV illumination. For consistency, the 0.50 mol% Eu³⁺-doped glass samples with side lengths of 2 cm, 4 cm, and 6 cm are hereafter referred to as L = 2 cm, L = 4 cm, and L = 6 cm, respectively. Compared to the measured η_edge value of 77.49% for L = 2 cm, the values for L = 4 cm, and L = 6 cm are 77.44%, and 77.30%, respectively, indicating that the intrinsic photon-guiding efficiency remains nearly constant with increasing waveguide dimensions. The η_abs(λ) measured for the larger samples are nearly identical to those of the reference L = 2 cm sample under the same optical conditions, confirming that all glass plates can be fabricated with high reproducibility and are readily scalable. Similarly, η_pl(λ) values exhibit only a negligible decrease with increasing device size (70.75% for L = 4 cm and 70.10% for L = 6 cm), indicating that reabsorption and scattering losses remain minimal even at larger dimensions and demonstrating the robustness of the material's optical quality during scaling. These parameters are then used to calculate η_ext(λ) and the values are 6.475% and 6.442%, respectively, which are closely aligned with the 6.483% obtained for the reference L = 2 cm sample. This consistency across all dimensions confirms effective photon confinement and robust waveguiding performance, independent of device size.

Fig. 2d illustrates the experimental setup used to measure distance-dependent edge emission for further evaluating the waveguiding ability of the samples. To minimize external interference, each sample is mounted in a custom PLA holder exposing only the front and rightmost faces. A collimated beam (1 × 6000 µm²) is directed perpendicularly onto the front surface, generating a high-intensity emission band along the horizontal excitation line. The fiber optic cable is aligned with the brightest point where this emission band reached the right edge of the sample. Throughout the measurements, the fiber tip remained in constant contact with the right edge of the sample to ensure consistent optical coupling. As the sample assembly is translated laterally to vary the separation between the illuminated region and the collection point, defined as the optical distance (L), this contact condition is strictly maintained.

Fig. 2e presents representative edge emission spectra for the largest sample (L = 6 cm), along with a schematic clarifying the optical distance term. With increasing L, emission intensity progressively declines, whereas the spectral profile exhibits minimal variation. To quantify this behavior, the integrated PL intensities are fitted using an exponential decay model:

I_L = I₀e^−βL (8)

where I_L is the PL intensity at distance L, and I₀ is the intensity at the shortest distance (5 mm). As shown in Fig. 2f, the fitted attenuation coefficient is β = 0.054 mm⁻¹. This value reflects cumulative optical losses during photon propagation, potentially arising from scattering, parasitic absorption, and/or undetected reabsorption effects.

Interestingly, despite the noticeable decline in PL intensity with increasing excitation distance, η_ext(λ) remains constant across all waveguide lengths. This apparent discrepancy arises from the nature of uniform solar illumination, which excites the entire surface area simultaneously. While localized excitation highlights photon attenuation along the propagation path, under full-area illumination, regions near the edges—where photon extraction is most efficient—dominate the overall output. Photons originating from more distant regions, though subject to higher losses, contribute only marginally due to their lower probability of reaching the edge via guided modes. Consequently, the proportional relationship between absorbed and collected photons remains stable across different device sizes. These results suggest that, even in the presence of moderate propagation losses, Eu³⁺-doped glass waveguides retain consistent optical performance under realistic solar conditions—supporting their suitability for scalable LSC applications.

3.3. Evaluation of optical, environmental and mechanical performance

In addition to collection, conversion and waveguiding performance, the aesthetic properties of the Eu³⁺-doped glasses are also examined to assess their suitability for architectural applications, where visual neutrality and high transparency are essential.

Fig. 3a presents the transmittance spectrum of the 0.50 mol% Eu³⁺-doped sample, which exhibits near-complete absorption below 320 nm and maintains high transmittance above 400 nm. The inset photograph, taken under natural daylight on the university campus, further illustrates the sample's remarkable transparency and colorless appearance. This optical clarity stems from the fact that Eu³⁺ absorption bands are primarily located in the UV region, thereby minimizing visible light attenuation and enabling both high visible transmittance and a visually neutral aesthetic.

Fig. 3 (a) Transmittance spectrum of the 0.50 mol% Eu³⁺-doped glass sample; the inset shows the 6 × 6 cm² sample under natural sunlight, demonstrating its optical clarity and homogeneity. (b) Temperature-dependent PL emission spectra, and (c) corresponding luminescence images at 50 °C and 200 °C. (d) Evolution of PL intensity over four consecutive heating–cooling cycles. (e) Transmittance and n_pl values monitored over 30-day immersion in deionized water. (f) PL intensity under continuous 393 nm excitation for 50 hours. (g) Vickers hardness values measured under varying loads; the inset shows representative indentations captured at 100× magnification.

Table 1 summarizes the AVT, CRI, and CIELAB chromaticity coordinates of Eu³⁺-undoped and doped glass samples. All compositions demonstrate excellent optical clarity (AVT ≥ 87.2%) and color fidelity (CRI ≥ 97), with the 0.50 mol% sample achieving the highest AVT of 89.42% and a CRI of 98. Importantly, the color appearance is highly natural, as quantified by its CIELAB coordinates (L* = 95.7, a* = 2.5, and b* = 7.2). The high lightness (L* > 95) and low chromaticity values (a* and b* close to zero) confirm a stable and visually unobtrusive appearance across all doping levels. These results confirm that Eu³⁺ incorporation does not compromise the visual aesthetics of the host glass, despite enhancing its luminescent functionality. The 0.50 mol% composition, already identified as photometrically optimal, also exhibits the most favorable visual characteristics, making it the most balanced and application-relevant formulation among all tested samples.

Table 1 AVT, CRI, and CIELAB coordinates of Eu³⁺-undoped and doped glass samples with doping concentrations of 0.25%, 0.50%, and 1.00%

Eu2O3 concentration (mol%)	AVT (%)	CRI	CIELAB (L*, a*, b*)
			L*	a*	b*
0.00%	87.69	97	95.0	1.5	10.4
0.25%	87.86	97	95.1	1.5	10.8
0.50%	89.42	98	95.7	2.5	7.2
1.00%	87.23	97	94.8	2.5	9.0

---

Maintaining visual transparency and aesthetic neutrality is equally critical for seamless integration into architectural environments. Thus, it is also essential to evaluate the environmental stability and durability of the selected composition. Accordingly, a comprehensive set of tests—including thermal stability, chemical durability, photostability, and mechanical hardness—is conducted on the 0.50 mol% Eu³⁺-doped glass to assess its practical viability for real-world deployment.

The thermal stability is assessed via temperature-dependent PL analysis in the range of 25–200 °C, as shown in Fig. 3b. Although a gradual decrease in PL intensity is observed with increasing temperature, the emission remains strong even at elevated temperatures, indicating good thermal tolerance. Notably, the emission spectrum shows no peak shift during heating, as evidenced by the stable position of the dominant 613 nm transition (inset of Fig. 3b). Fig. 3c presents photographs of the sample positioned within the heating accessory of the spectrofluorometer at 50 °C and 200 °C. No visible changes in emission color or intensity are observed, supporting the spectroscopic findings. To further probe thermal resilience, the PL response is monitored over four heating–cooling cycles. Fig. 3d displays the maximum emission intensities at 613 nm recorded during each cycle. While a minor decrease in intensity is noted upon cooling, the emission consistently returns to its original shape and strength across cycles. This reversible behavior confirms the excellent thermal stability and structural integrity of the glass matrix under cyclic temperature variation—an essential feature for outdoor LSC applications.

The chemical durability is evaluated by immersing 0.50 mol% Eu³⁺-doped glass in deionized water for 30 days. During this period, both η_pl and optical transmittance are monitored at three-day intervals, as shown in Fig. 3e. The results indicate that the luminescence performance of Eu³⁺ ions remains fully preserved, with no degradation in η_pl observed throughout the test period. Furthermore, prolonged immersion has no measurable impact on the transmittance of the glass, confirming its excellent chemical durability in aqueous environments. No signs of hazing, fogging, or surface degradation—typically associated with glass corrosion—are observed during or after the test period.

Fig. 3f presents the results of a photostability test in which the glass sample is subjected to continuous 393 nm excitation for 50 hours. Emission measurements are recorded hourly during the first 10 hours, and subsequently at 10-hour intervals. Although minor fluctuations are observed in the short-term data, the overall PL intensity remains stable throughout the test duration. This consistency confirms that prolonged UV exposure does not induce defect formation or degradation in the Eu³⁺-doped glass matrix, demonstrating its excellent photostability under extended illumination.

Fig. 3g presents the Vickers hardness test results under varying loads. For each load, five measurements are performed, and the average values are reported. As expected, a slight increase in hardness is observed with increasing applied load, reaching a maximum of 5.6 GPa at 100 gf. This high hardness highlights the mechanical strength of the borosilicate glass matrix, confirming its structural robustness and suitability for practical deployment in demanding environments.

These stability tests confirm the intrinsic chemical and photochemical robustness of the Eu³⁺-doped glasses, demonstrating their potential as a durable luminescent component for BIPV-integrated LSC systems. However, it should be emphasized that the present device represents a proof-of-concept configuration rather than a fully assembled and encapsulated module. Accordingly, future studies are needed focusing on the fabrication of monolithic, fully assembled LSC devices that can undergo industry-standard qualification procedures.

3.4. LSC device construction study

Fig. 4a shows the system used to evaluate photovoltaic performance under AM1.5G illumination conditions. In this setup, waveguides are combined with resized commercial monofacial PERC silicon solar cells to form LSC devices. The figure depicts both the schematic arrangement and a photograph of both the system and the cut-cells (8.0 × 1.0 cm²). The waveguides are mounted in black PLA holders that shield the bottom and three lateral surfaces to suppress stray light. The waveguides were directly coupled to the solar cells without a refractive index matching layer to avoid introducing additional engineering parameters at the interface. This simplified setup ensures that the measured performance directly reflects the intrinsic optical properties of the glass samples under reproducible conditions. The emitting edge of the glass waveguide is precisely aligned with the active region of the solar cell, and the contact area is confined by the edge of the PLA holder, as shown in the schematic. The solar cell is seated in a rail-like groove within the holder, positioned approximately 1 mm recessed from the outer surface. This recessed placement effectively prevents direct illumination of the cell by ambient or incident light. To further minimize ambient light interference, the remaining cell width not covered by the waveguide is masked using a black, highly absorptive material.

Fig. 4 (a) Schematic and photograph of the custom measurement setup integrating the waveguide and cut cells, shown for the L = 2 cm device under AM1.5G illumination. (b) J–V curves of LSC devices with waveguide lengths of 2, 4, and 6 cm, along with the reference cut-cell. (c) External quantum efficiency of the LSC devices with varying lengths. (d) EQE_LSC(λ) of the 6 cm device alongside the PL spectrum, cut-cell EQE, and AM1.5G solar spectrum. EQE_LSC(λ) and EQE are expressed in %, while PL and AM1.5G spectra are shown in arbitrary units.

Fig. 4b shows the current density–voltage (J–V) characteristics of the cut silicon solar cell measured under standard 100 mW cm⁻² AM1.5G illumination. The measured short-circuit current density J^meas_sc is calculated by taking the illuminated top surface as the photoactive area, since photon collection occurs through this surface in the waveguide configuration. Accordingly, J^meas_sc decreases from 0.687 mA cm⁻² to 0.124 mA cm⁻² with increasing device size. This reduction is attributed to the geometrical gain factor (G) of the LSC, which is defined as the ratio of the PV collection area to the device aperture area. As G increases with device size, the normalized current density diminishes—even though more photons are guided to the PV interface.

Table 2 presents the photovoltaic parameters (V_oc, I_sc, FF, and J_sc) for the L = 2, 4, and 6 cm LSC devices, showing both calculated and experimentally measured values, normalized to four edges. As waveguide length increases, a moderate increase in V_oc is observed, from 463 to 525 mV, reflecting the higher total photon flux reaching the solar cell interface. I_sc increases from 2.749 mA to 4.472 mA, due to enhanced photon collection and waveguiding across the larger illuminated area. These trends indicate that scaling up the LSC dimensions can improve absolute power output—a favorable feature for BIPV applications. The FF remains constant at about ∼67%, indicating stable collection charge efficiency across all device sizes. To enable a fair comparison with a fully edge-coupled configuration, the term 4 × J^meas_sc represents the total current density, assuming that all four edges of the LSC are coupled to solar cells instead of the single-edge configuration used in this study.

Table 2 Photovoltaic parameters of the L = 2, 4, and 6 cm LSC devices, including both calculated and measured values normalized to four edges

L (cm)	Voc (mV)	Isc (mA)	J^meassc (mA cm^−2)	J^calcsc (mA cm^−2)	4 × J^meassc (mA cm^−2)	4 × J^calcsc (mA cm^−2)	FF (%)
2	463	2.749	0.687	0.598	2.749	2.392	66.85
4	493	3.165	0.198	0.299	0.792	1.196	67.14
6	525	4.472	0.124	0.199	0.496	0.796	67.11

---

The external quantum efficiency of the LSC devices, EQE_LSC(λ), is calculated through a reported mathematical model⁴² based on η_ext(λ), primarily to approximate the system's wavelength-dependent photon collection behaviour:

where denotes the external quantum efficiency of the edge-coupled silicon cell (measured to be ∼95%) around the Eu³⁺ emission band. Fig. 4c presents the EQE_LSC(λ) spectra for the LSC devices with dimensions of L = 2, 4, and 6 cm, plotted alongside the experimental PL excitation spectrum of the glass doped with 0.50 mol% Eu₂O₃. The spectral profiles exhibit strong correspondence with the PL excitation spectrum. However, minor deviations are observed, particularly due to the absence of features corresponding to the PLE peaks at approximately 318 nm and 576 nm. The former is explained by η_pl being defined as zero in the non-emissive UV region dominated by host absorption. The latter arises from the intrinsically weak transition, whose contribution lies below the sensitivity of the estimation method.

The EQE_LSC(λ) is then used to calculate the short-circuit current density, J_sc, as expressed by the following relationship:⁴²

where e is the elementary charge, and ϕ_AM1.5G(λ) is the incident photon flux. The calculated J^calc_sc values are 0.598, 0.299, and 0.199 mA cm⁻² for the L = 2, 4, and 6 cm devices, respectively. The model shows excellent agreement for the L = 2 cm device and remains consistent for larger ones, with the divergence at greater sizes caused by increased optical losses at the cut-cell–waveguide interface.

Fig. 4d shows the EQE_LSC(λ) of the L = 6 cm device together with the PL spectrum, reference cell EQE, and AM1.5G solar spectrum. The LSC exhibits strong UV-blue activity, efficiently converting high-energy photons, while its emission band aligns with the cut-cell's peak quantum efficiency, confirming excellent spectral matching and reliable photovoltaic performance under solar illumination.

Table 3 provides a comparative overview of the 0.50 mol% Eu³⁺-doped glass-based LSCs developed in this study, alongside a broad selection of previously reported Eu³⁺-based LSC systems. In addition to reporting the PCE values for three device sizes (L = 2, 4, and 6 cm), the table compiles key literature data, including LSC type (planar or cylindrical), luminescent species, host matrix, processing method (film, fiber, or bulk), device size, and measurement conditions. Although some literature sources report optical efficiency (OE), this metric is not included in this study due to the lack of a standardized definition or consistent methodology. In current literature, OE is calculated using differing assumptions, equations, and boundary conditions, leading to significant variability and limiting direct comparability. To ensure clarity and consistency in performance evaluation, PCE is selected as the primary metric, as it provides a more robust and standardized measure of LSC performance.⁴³ PCE is calculated using the following equation, where all electrical parameters correspond to the LSC device, and P_in represents the standard incident power density of 100 mW cm⁻² under AM1.5G illumination:

Table 3 Overview of Eu³⁺-based LSC studies, categorized by LSC type, luminophore characteristics, host materials, processing method, device size, η_pl, OE, and PCE (the table summarizes performance variations, adherence to the consensus statement,⁴⁴ and peak or representative PCE values, with full citations provided in the reference section)

Luminophore	Host material	Processing method	Device size (cm)	ηpl (%)	OE (%)	PCE (%)	Deviations	Year	Ref.
a Studies classified as “in agreement” with the consensus statement^33 adhere to standardized LSC measurement protocols. In contrast, deviations include one or more of the following: (i) lack of edge covering, (ii) use of a reflective background/edge covering, and (iii) insufficient experimental detail. Studies that do not report any PCE values are marked with (–), as the comparison is based solely on reported PCE data.
Eu(tta)3(Phen)	PVB	Film	7.8 × 7.8 × 0.3	44	—	0.0499	(iii)	2011	14
Eu(tta)3(TPPO)2	PMMA	Film	10 × 10 × 0.3	73	—	0.28	(iii)	2011	20
Eu^3+@diureido-bipyridine	Bridged silsesquioxane (BS)	Film	∼2.5 × 2.5 × 0.8	34	1.7	—	—	2011	28
Eu^3+@diureido-bipyridine	Modified organosilane	Film	2.5 × 2.5 × 0.08	8	1.2	—	—	2012	29
Eu(tta)3(Phen)	Tri-ureasil 5000	Film	—	63	—	—	—	2013	21
[Eu(btfa)3(MeOH)2]2bpta2	Tri-ureasil 5000	Film	2.5 × 2.5 × 1.0	27	3.2	0.07	(ii)	2014	23
Eu(TTA)3phen	Parylene	Film	5 × 5 × 0.5	—	2.47	0.19	^a	2015	24
Eu^3+@diureido-bipyridine	BS	Film	2.5 × 2.5 × 0.14	23	0.43	0.3	(iii)	2015	30
Eu^3+:ionossilica	PMMA	Film	7.5 × 2.0 × 0.1	31	0.34	0.0019	(ii)	2019	31
Eu(2mCND)3	Di-ureasil	Film	5 × 5 × 0.4	69	0.51	0.138	(i, ii)	2022	25
Lu red (commercial)	Acrylic resin	Film	5 × 5 × 0.3	20–36	—	0.70	^a	2022	19
Eu^3+ tetrakis (β-diketonate)	PMMA	Film	3 × 3 × 0.4	63	5.40	0.096	(i)	2022	26
[Eu2LF4](NEt4)2	PMMA	Bulk	5 × 5 × 0.27	50	—	0.124	^a	2024	27
Eu^3+	Glass	Bulk	2 × 2 × 0.25	78	—	0.213 × 4	^a	2025	This work
			4 × 4 × 0.25			0.066 × 4
			6 × 6 × 0.25			0.041× 4

---

The resulting LSC prototypes exhibit excellent optical transparency and color neutrality, yielding PCE values of 0.213%, 0.066%, and 0.041% for device lengths of 2, 4, and 6 cm, respectively, coupled to a single PV cell mounted along one edge. For a fair comparison with fully edge-coupled configurations, these values can be scaled by a factor of four to represent a device in which all four edges are interfaced with PV cells. Under this assumption, the corresponding four-edge–equivalent PCEs (4 × PCEs) are 0.852%, 0.264%, and 0.164% for the 2, 4, and 6 cm devices, respectively comparable to or surpassing the performance of many previously reported Eu³⁺-based systems.

The observed decrease in PCE can be primarily explained by the geometric gain factor (G), which is inversely proportional to the device length (G ∝ 1/L). This geometric scaling alone accounts for a threefold efficiency reduction when the device dimension increases from 2 cm to 6 cm. However, the experimental results reveal a slightly larger decrease (∼4.8-fold), suggesting the presence of additional loss mechanisms. The slightly larger-than-expected reduction is attributed to minor interfacial losses at the glass–PV interface, caused by small air gaps in the proof-of-concept assembly. These losses are not intrinsic to the Eu³⁺-doped glass and could be minimized using standard index-matching techniques in future scalable devices.

A review of the literature indicates that most Eu³⁺-based LSCs are fabricated using polymeric matrices such as PMMA, PVB, or various resin systems, typically in film or bulk configurations. While these materials offer ease of processing and good initial optical quality, they are limited by poor long-term stability, low UV resistance, and vulnerability to thermal and mechanical degradation—factors that significantly hinder their suitability for outdoor applications. Furthermore, despite the incorporation of advanced Eu³⁺ complexes or co-doping strategies, many reported PCEs remain extremely low, with some as low as 0.001%. Even in systems with high photoluminescence quantum yields (η_pl > 70%), overall device performance is frequently constrained by inefficient photon waveguiding or limited compatibility with scalable, integrated device architectures.

It is essential to acknowledge that variations in experimental methodologies throughout literature interrupt direct comparison of reported PCE values. Table 3 shows that many studies lack essential information regarding standard testing conditions (AM1.5G, 100 mW cm⁻²) or omit critical measurement protocols such as spectral mismatch correction, optical coupling configuration, and edge masking. In several cases, reported efficiencies are obtained from non-standard or extrapolated measurement setups, further increasing variability. Moreover, the number of PV-coupled edges differs significantly across the literature, making direct comparison even more challenging. Such methodological discrepancies can lead to over- or underestimation of performance metrics, emphasizing the need for standardized reporting to enable reliable benchmarking in LSC research. By contrast, the present study follows established characterization protocols, ensuring a more accurate and reproducible assessment of device-level performance.

Despite the diversity in reporting approaches across the literature, a few studies provide methodologically transparent benchmarks. For example, Bertozzi et al.¹⁹ report a peak PCE of 0.70% for a film-type LSC using a commercial red dye in an acrylic resin, while Motta et al.²⁷ achieve 0.124% using supramolecular Eu³⁺-based cages as hybrid luminophores—both values corresponding to fully four-edge–coupled configurations. Collectively, these studies represent important advances in the development of Eu³⁺-based LSCs, particularly in improving optical performance through tailored material and device strategies.

Overall, while the literature demonstrates important advances in Eu³⁺-based LSCs, most reported systems still face intrinsic limitations under realistic conditions due to their polymer-based designs. In this context, Eu³⁺-doped glass-based LSC devices developed in this study offer distinct advantages, including high optical clarity and visible transparency, along with UV-cut functionality and superior chemical, thermal, and mechanical robustness. These attributes directly address the stability and integration challenges that hinder prior systems, positioning the glass-based platform as a durable and regulation-compliant material option for scalable, building-integrated LSC applications.

4. Conclusions

This study demonstrates, for the first time, the feasibility of employing Eu³⁺-doped glasses as both luminescent media and optical waveguides in fully inorganic LSCs. Transparent and color-neutral planar devices (2.5 mm thick; 2 × 2 to 6 × 6 cm²) exhibit a consistent external photon efficiency of approximately 6.4% under AM1.5G solar illumination, indicating efficient photon collection and guiding within the glass matrix—even in the presence of distance-dependent attenuation under localized excitation. 4 × PCEs of 0.852%, 0.264%, and 0.164% for increasing device sizes confirm the scalability and performance retention of the proposed glass-based architecture. Compared to existing Eu³⁺-based LSCs fabricated from polymeric or hybrid matrices, the Eu³⁺-doped glass platform offers markedly enhanced environmental stability, UV resistance, and compatibility with scalable manufacturing. These findings not only validate the integration of rare-earth-doped glasses in LSC technologies but also establish a new materials paradigm for robust, durable, and architecturally integrable BIPV applications.

Author contributions

Emre İlter: investigation, validation, writing – original draft, and visualization. Utku Ekim: investigation, validation, and writing – review & editing. Harun Samet Çelik: investigation, validation, and writing – original draft. Miray Çelikbilek Ersundu: methodology, resources, writing – review & editing, supervision, project administration, and funding acquisition. Ali Erçin Ersundu: conceptualization, methodology, resources, writing – original draft, supervision, project administration, and funding acquisition.

Conflicts of interest

The authors declare no competing financial interests.

Data availability

The data supporting this article are available within the main article.

Acknowledgements

The authors gratefully acknowledge the Scientific and Technological Research Council of Turkey (TÜBİTAK) for financial support under project numbers 221M549 and 223N020, as well as the Yildiz Technical University Scientific Research Projects Coordination Unit for funding through projects FYL-2025-6869, FBA-2024-6108, and FBA-2025-6788. The authors thank Assoc. Prof. Dr Selçuk Yerci (ODTÜ-GÜNAM – The Center for Solar Energy Research and Applications) and Konstantin Tsoi for kindly providing the silicon solar cells used in this study, along with their characterization data. Finally, Ali Erçin Ersundu acknowledges support from the Turkish Academy of Sciences (TÜBA) through the Outstanding Young Scientists Award (GEBİP) program in recognition of his studies on luminescent solar concentrators.

References

1. IEA, Empowering Cities for a Net Zero Future, Paris, 2021.
2. R. van Heerden, O. Y. Edelenbosch, V. Daioglou, T. Le Gallic, L. B. Baptista, A. Di Bella, F. Pietro Colelli, J. Emmerling, P. Fragkos, R. Hasse, J. Hoppe, P. Kishimoto, F. Leblanc, J. Lefèvre, G. Luderer, G. Marangoni, A. Mastrucci, H. Pettifor, R. Pietzcker, P. Rochedo, B. van Ruijven, R. Schaeffer, C. Wilson, S. Yeh, E. Zisarou and D. van Vuuren, Nat. Energy, 2025, 10, 380–394.
3. D. Gielen, F. Boshell, D. Saygin, M. D. Bazilian, N. Wagner and R. Gorini, Energy Strategy Rev., 2019, 24, 38–50.
4. C. Xiang, B. S. Matusiak, A. Røyset and T. Kolås, Sol. Energy, 2021, 224, 1222–1246.
5. K. Lee, H.-D. Um, D. Choi, J. Park, N. Kim, H. Kim and K. Seo, Cell Rep. Phys. Sci., 2020, 1, 100143.
6. H. Lee and H.-J. Song, WIREs Energy Environ., 2021, 10, e403.
7. G. Figueiredo, S. F. H. Correia, L. Fu, V. de Zea Bermudez, A. N. C. Neto, P. S. Andre and R. A. S. Ferreira, Luminescent solar concentrators: Current and future applications in smart cities, Handbook on the Physics and Chemistry of Rare Earths, Elseiver, 2024, vol. 66, pp. 51–123.
8. M. G. Debije and V. A. Rajkumar, Sol. Energy, 2015, 122, 334–340.
9. J. Lin, L. Wang, Q. Jing and H. Zhao, Chem. Eng. J., 2024, 481, 148441.
10. J. Lin, L. Wang, X. Meng, W. Li, N. Ren, L. Tao, J. Xiao, Q. Jing, Y. Song and H. Zhao, Appl. Phys. Lett., 2025, 126, 043904.
11. S. Yano, Y. Iso and T. Isobe, ACS Appl. Nano Mater., 2026, 9, 543–550.
12. B. S. Richards and I. A. Howard, Energy Environ. Sci., 2023, 16, 3214–3239.
13. C. Yang and R. R. Lunt, Adv. Opt. Mater., 2017, 5, 1600851.
14. X. Wang, T. Wang, X. Tian, L. Wang, W. Wu, Y. Luo and Q. Zhang, Sol. Energy, 2011, 85, 2179–2184.
15. S. F. H. Correia, P. P. Lima, P. S. André, M. R. S. Ferreira and L. A. D. Carlos, Sol. Energy Mater. Sol. Cells, 2015, 138, 51–57.
16. S. F. H. Correia, P. P. Lima, E. Pecoraro, S. J. L. Ribeiro, P. S. André, R. A. S. Ferreira and L. D. Carlos, Prog. Photovoltaics Res. Appl., 2016, 24, 1178–1193.
17. I. Parola, M. A. Illarramendi, F. Jakobs, J. Kielhorn, D. Zaremba, H.-H. Johannes and J. Zubia, Polymers, 2019, 11, 1187.
18. S. F. H. Correia, A. R. Frias, L. Fu, R. Rondão, E. Pecoraro, S. J. L. Ribeiro, P. S. André, R. A. S. Ferreira and L. D. Carlos, Adv. Sustainable Syst., 2018, 2, 1800002.
19. A. F. Bertozzi, A. Picchi and A. Pucci, Macromol. Chem. Phys., 2023, 224, 2200392.
20. T. Wang, J. Zhang, W. Ma, Y. Luo, L. Wang, Z. Hu, W. Wu, X. Wang, G. Zou and Q. Zhang, Sol. Energy, 2011, 85, 2571–2579.
21. M. M. Nolasco, P. M. Vaz, V. T. Freitas, P. P. Lima, P. S. André, R. A. S. Ferreira, P. D. Vaz, P. Ribeiro-Claro and L. D. Carlos, J. Mater. Chem. A, 2013, 1, 7339.
22. I. S. Grigoryev, L. G. Klapshina, S. A. Lermontova, V. V. Semenov, V. M. Treushnikov, V. V. Treushnikov, B. A. Bushuk, S. Clement and W. E. Douglas, Nanotechnol. Russ., 2012, 7, 492–498.
23. S. F. H. Correia, V. de Zea Bermudez, S. J. L. Ribeiro, P. S. André, R. A. S. Ferreira and L. D. Carlos, J. Mater. Chem. A, 2014, 2, 5580–5596.
24. M. Tonezzer, G. Maggioni, A. Campagnaro, S. Carturan, A. Quaranta, M. della Pirriera and D. Gutierrez Tauste, Prog. Photovoltaics Res. Appl., 2015, 23, 1037–1044.
25. Y. Wang, Y. Liu, G. Xie, J. Chen, P. Li, Y. Zhang and H. Li, ACS Appl. Mater. Interfaces, 2022, 14, 5951–5958.
26. Y. Wang, G. Xie, J. Chen, X. Zhang, C. Chen, J. Yin and H. Li, J. Mater. Chem. C, 2022, 10, 11924–11930.
27. I. Motta, G. Bottaro, M. Rando, M. Rancan, R. Seraglia and L. Armelao, J. Mater. Chem. A, 2024, 12, 22516–22527.
28. J. Graffion, X. Cattoën, M. Wong Chi Man, V. R. Fernandes, P. S. André, R. A. S. Ferreira and L. D. Carlos, Chem. Mater., 2011, 23, 4773–4782.
29. J. Graffion, A. M. Cojocariu, X. Cattoën, R. A. S. Ferreira, V. R. Fernandes, P. S. André, L. D. Carlos, M. Wong Chi Man and J. R. Bartlett, J. Mater. Chem., 2012, 22, 13279.
30. V. T. Freitas, L. Fu, A. M. Cojocariu, X. Cattoën, J. R. Bartlett, R. Le Parc, J.-L. Bantignies, M. Wong Chi Man, P. S. André, R. A. S. Ferreira and L. D. Carlos, ACS Appl. Mater. Interfaces, 2015, 7, 8770–8778.
31. A. R. Frias, M. A. Cardoso, A. R. N. Bastos, S. F. H. Correia, P. S. André, L. D. Carlos, V. de Zea Bermudez and R. A. S. Ferreira, Energies, 2019, 12, 451.
32. M. Riahinezhad, M. Hallman and J.-F. Masson, Buildings, 2021, 11, 299.
33. G. Griffini, M. Levi and S. Turri, Sol. Energy Mater. Sol. Cells, 2013, 118, 36–42.
34. T. N. L. Tran, A. Chiasera, A. Lukowiak and M. Ferrari, Materials, 2022, 15, 1847.
35. J. E. Shelby, Introduction to Glass Science and Technology, The Royal Society of Chemistry, 2005.
36. J. Zhao, L. Huang, T. Liang, S. Zhao and S. Xu, J. Lumin., 2019, 205, 342–345.
37. J. Zhang, G. Cai, W. Wang, L. Ma, X. Wang and Z. Jin, Inorg. Chem., 2020, 59, 2241–2247.
38. M. Wei, F. P. G. de Arquer, G. Walters, Z. Yang, L. N. Quan, Y. Kim, R. Sabatini, R. Quintero-Bermudez, L. Gao, J. Z. Fan, F. Fan, A. Gold-Parker, M. F. Toney and E. H. Sargent, Nat. Energy, 2019, 4, 197–205.
39. K. Wu, H. Li and V. I. Klimov, Nat. Photonics, 2018, 12, 105–110.
40. H. Li, K. Wu, J. Lim, H.-J. Song and V. I. Klimov, Nat. Energy, 2016, 1, 16157.
41. V. I. Klimov, T. A. Baker, J. Lim, K. A. Velizhanin and H. McDaniel, ACS Photonics, 2016, 3, 1138–1148.
42. C. K. Herrera, A. Vahdani, C. Yang, M. Bates, S. Y. Lunt, B. Borhan and R. R. Lunt, ACS Photonics, 2023, 10, 3195–3202.
43. C. Yang, D. Liu and R. R. Lunt, Joule, 2019, 3, 2871–2876.
44. C. Yang, H. A. Atwater, M. A. Baldo, D. Baran, C. J. Barile, M. C. Barr, M. Bates, M. G. Bawendi, M. R. Bergren, B. Borhan, C. J. Brabec, S. Brovelli, V. Bulović, P. Ceroni, M. G. Debije, J.-M. Delgado-Sanchez, W.-J. Dong, P. M. Duxbury, R. C. Evans, S. R. Forrest, D. R. Gamelin, N. C. Giebink, X. Gong, G. Griffini, F. Guo, C. K. Herrera, A. W. Y. Ho-Baillie, R. J. Holmes, S.-K. Hong, T. Kirchartz, B. G. Levine, H. Li, Y. Li, D. Liu, M. A. Loi, C. K. Luscombe, N. S. Makarov, F. Mateen, R. Mazzaro, H. McDaniel, M. D. McGehee, F. Meinardi, A. Menéndez-Velázquez, J. Min, D. B. Mitzi, M. Moemeni, J. H. Moon, A. Nattestad, M. K. Nazeeruddin, A. F. Nogueira, U. W. Paetzold, D. L. Patrick, A. Pucci, B. P. Rand, E. Reichmanis, B. S. Richards, J. Roncali, F. Rosei, T. W. Schmidt, F. So, C.-C. Tu, A. Vahdani, W. G. J. H. M. van Sark, R. Verduzco, A. Vomiero, W. W. H. Wong, K. Wu, H.-L. Yip, X. Zhang, H. Zhao and R. R. Lunt, Joule, 2022, 6, 8–15.

---