Multi-wavelength UV Upconversion in Lanthanides assisted by Photonic Crystals

Technical Summary: Multi-wavelength UV Upconversion in Lanthanides assisted by Photonic Crystals

Problem Statement
Upconversion luminescence (UCL) involves the sequential absorption of low-energy photons to emit higher-energy photons. While extensively studied for converting near-infrared (NIR) to visible light in lanthanides, the conversion of broad excitations (specifically NIR and visible) to the ultra-violet (UV) range remains underexplored. This conversion is critical for applications such as solar-to-fuel conversion and environmental remediation. However, UCL efficiency in lanthanides is fundamentally limited by the low absorption cross-sections of 4f-4f transitions ($\sigma_{abs} \sim 10^{-20} \text{ cm}^2$), cross-relaxations, and surface quenching. Specifically, the mechanism for generating UV emission from Yb$^{3+}$-Tm$^{3+}$ co-doped systems under solar-like irradiance relies on a multi-wavelength excitation scheme (NIR + visible) involving excited-state absorption (ESA), which is inherently weak without enhancement.

Methodology
The authors engineered a photonic crystal (PhC) slab to assist this multi-wavelength upconversion mechanism by manipulating the local photonic density of states (LDOS). The approach involved:

1. Material Design: A thin film stack was fabricated consisting of a 95 nm Y$_2$O$_3$ layer doped with 7.5% Yb$^{3+}$ and 0.5% Tm$^{3+}$ (deposited via Pulsed Laser Deposition) on top of a 144 nm SiN$_x$ layer (deposited via PECVD). The stoichiometry and thickness were optimized to ensure transparency from UV to NIR while maximizing the refractive index contrast.
2. Photonic Crystal Engineering: The SiN$_x$ layer was patterned with a square lattice of air holes using electron-beam lithography and inductively coupled plasma reactive ion etching (ICP-RIE). The lattice period ($p$) was scanned between 240 nm and 280 nm.
3. Resonance Tuning: The geometric parameters were designed to create "slow-light" Bloch modes at the $\Gamma$-point (normal incidence). These modes were tuned to overlap with the Tm$^{3+}$ excited-state absorption transition ($^3$F$_4 \rightarrow ^1$D$_2$) at visible wavelengths (450 nm) to enhance absorption, while simultaneously creating Bloch modes in the UV emission range (370 nm) to enhance light extraction.
4. Characterization: The team measured UCL power laws to confirm the mechanism, performed reflectivity measurements to map the photonic band structures, and compared the performance of 22 different PhC structures against an unpatterned reference stack.

Key Contributions

• LDOS Engineering for Multi-wavelength Regimes: Unlike previous studies focusing on single-wavelength (NIR) excitation, this work demonstrates LDOS engineering specifically for a multi-wavelength regime (NIR + visible) to drive UV upconversion.
• Decoupling of Contributions: The study quantitatively separates the enhancement factors into two distinct contributions: the enhancement of visible light absorption (via slow-light resonance) and the enhancement of UV light extraction (via Bloch modes).
• Fabrication of Low-Loss Structures: The design utilizes a transparent host matrix (Y$_2$O$_3$/SiN$_x$) to avoid parasitic absorption losses often found in plasmonic or opal-based approaches.

Results

• Enhancement Factor: The optimized PhC structure (Structure D, with a lattice period of ~267 nm) achieved a 28.1 $\pm$ 3.0-fold enhancement in integrated UV UCL intensity compared to the unpatterned reference under combined NIR (995 nm) and visible (447 nm) excitation.
• Mechanism Confirmation: Power law measurements confirmed the upconversion mechanism follows the multi-wavelength regime where the UCL intensity is proportional to the first power of both NIR and visible irradiances ($n_{NIR} = n_{vis} = 1$), validating the involvement of one NIR photon and one visible photon.
• Quantified Contributions:
  • UV Extraction: Under NIR-only excitation (where no visible ESA occurs), the PhC provided a 2.71 $\pm$ 0.5-fold enhancement, attributed solely to improved UV extraction via Bloch modes.
  • Visible Absorption: By subtracting the extraction factor from the total enhancement, the authors deduced that the slow-light resonance enhanced the visible excited-state absorption (ESA) by a factor of 10.4 $\pm$ 2.2. This corresponds to a local irradiance increase of the same magnitude at the visible wavelength.
• Sensitivity to Lattice Parameters: The enhancement was highly sensitive to the lattice period. Structures where the slow-light resonance (specifically the TE$_0$-like mode) spectrally and angularly overlapped with the Tm$^{3+}$ absorption cross-section and the visible laser excitation showed the highest enhancement. Structures with poor overlap (e.g., Structure C) showed significantly lower enhancement.
• Simulation Agreement: Measured band structures and reflectivity spectra showed excellent agreement with Rigorous Coupled-Wave Analysis (RCWA) simulations, confirming the presence of slow-light resonances at the $\Gamma$-point in the visible range and Bloch modes in the UV range.

Significance
The paper claims this is the first demonstration of multi-wavelength excitation assisted by LDOS engineering in lanthanides. The significance lies in overcoming the intrinsic low absorption cross-sections of lanthanides for the specific transition required for UV upconversion ($^3$F$_4 \rightarrow ^1$D$_2$). By utilizing slow-light resonances to trap visible light and Bloch modes to extract UV light, the authors achieved a substantial enhancement without introducing parasitic losses. The authors suggest that this methodology could be applied to other transitions or systems, including fluorinated host matrices or different sensitizer/activator pairs, and that the design principles could be scaled using large-area lithography techniques (e.g., nanoimprint) for potential solar-based applications. Additionally, the enhancement of ESA via PhCs could be relevant for reducing thresholds in photon-avalanching regimes.