Mid-Infrared Thermal Radiation Harvesting using Uncooled Narrow Bandgap GeSn Thermophotovoltaic cell

1. Problem Statement

Thermophotovoltaic (TPV) technology offers a promising solution for converting thermal radiation (specifically in the 500–2000 K range) into electricity, with applications in industrial waste heat recovery, aerospace power management, and portable energy generation. However, practical deployment of mid-wave infrared (MWIR) TPV systems faces significant material challenges:

• Current Limitations: High-performance TPV devices rely on III-V semiconductors (e.g., InAs, GaSb, InGaAs). While efficient, these materials are constrained by high manufacturing costs, small substrate sizes, and incompatibility with standard silicon processing, limiting scalability for large-area applications.
• The Gap: Group-IV GeSn alloys are a promising alternative due to their silicon compatibility and tunable direct bandgaps in the MWIR. While theoretical models have predicted their potential, no experimental demonstration of a GeSn-based TPV device had been reported prior to this work.

2. Methodology

The authors fabricated and characterized proof-of-concept GeSn TPV diodes, benchmarking them against commercial III-V devices.

• Material Growth:

  • Substrate: 4-inch Si(100).
  • Structure: Vertical p-i-n heterostructure grown via Reduced Pressure Chemical Vapor Deposition (RPCVD).
  • Layers: A four-layer graded GeSn buffer topped with a p-i-n stack:
    • p-type: Ge$_{0.94}$Sn$_{0.06}$
    • i-type (active): Ge$_{0.91}$Sn$_{0.09}$ (direct bandgap)
    • n-type: Ge$_{0.95}$Sn$_{0.05}$
  • Doping: Active carrier concentrations > $10^{19}$ cm$^{-3}$ achieved using diborane and arsine.
  • Quality: The intrinsic (i) and n-layers showed high crystalline quality, though the p-layer and buffer contained dislocations near the Si/Ge interface.

• Device Fabrication:

  • Geometry: Circular mesa vertical p-i-n diodes with a 1 mm active area diameter.
  • Process: Included surface cleaning, anisotropic ICP-RIE etching to form a double-mesa structure for isolation, SiO$_2$ passivation, and Ti/Au/Ag/Au metal contacts.
  • Design Choice: The circular mesa design was chosen to minimize fabrication uncertainties and focus on intrinsic junction behavior rather than maximizing efficiency via complex gridlines.

• Experimental Setup:

  • Illumination Sources:
    1. Monochromatic: 2.3 $\mu$m laser ($P_{in}$ = 5.14 mW).
    2. Broadband Thermal: SiC Globar emitter at $\sim$1500 K (gray-body spectrum 0.5–9.0 $\mu$m).
  • Comparison: Devices were tested alongside commercial InAs and extended-InGaAs photodiodes of identical geometry (1 mm diameter).
  • Measurements: Current-Voltage (I-V) characteristics, spectral responsivity (FTIR), and output power density under dark, laser, and thermal conditions.

• Modeling:

  • A 1D Poisson-drift-diffusion model was used to simulate intrinsic performance limits.
  • The model incorporated experimentally calibrated emitter emissivity and accounted for Shockley-Read-Hall (SRH), Auger, and surface recombination mechanisms.

3. Key Contributions

• First Experimental Demonstration: This work provides the first proof-of-concept experimental validation of a GeSn-based TPV diode for MWIR thermal energy harvesting.
• Benchmarking: Direct comparison of GeSn performance against state-of-the-art InAs and ext-InGaAs devices under identical thermal and laser illumination conditions.
• Performance Gap Analysis: Quantitative separation of experimental limitations (defects, transport) from intrinsic material potential using calibrated numerical modeling.
• Scalability Pathway: Demonstrates a silicon-compatible platform that could eventually replace costly III-V materials for large-area TPV applications.

4. Results

Experimental Performance (300 K):

• Responsivity: The GeSn device achieved a peak responsivity of ~0.2 A/W at 1.7 $\mu$m, with a cutoff near 2.6 $\mu$m. This is lower than the InAs (1 A/W) and ext-InGaAs (~1.52 A/W) counterparts.
• Dark Current: GeSn devices exhibited higher dark currents (0.28 $\mu$A at 0 V) compared to ext-InGaAs (34.5 nA), resulting in lower rectification ratios.
• Power Output (SiC Emitter, ~1500 K):
  • GeSn: ~0.41 mW/cm$^2$.
  • InAs: ~4.3 mW/cm$^2$.
  • ext-InGaAs: ~64 mW/cm$^2$.
• Laser Illumination (2.3 $\mu$m): GeSn produced an output power density of 0.27 mW/cm$^2$, significantly lower than the commercial devices but confirming functional photo-response.

Simulation vs. Experiment:

• Intrinsic Potential: Theoretical modeling predicted a peak power density of ~1.37 W/cm$^2$ for an ideal GeSn device under the same thermal conditions.
• The Discrepancy: The experimental output (0.41 mW/cm$^2$) is roughly 3,000 times lower than the simulated potential.
• Root Cause: The massive performance gap is attributed to defect-assisted recombination (specifically SRH recombination due to misfit dislocations and point defects) and transport losses, rather than fundamental material limitations.
• Sensitivity: Simulations showed that reducing the SRH carrier lifetime from a "Ge-like" baseline ($2 \times 10^{-6}$ s) to a defect-dominated regime ($10^{-10}$ s) causes the open-circuit voltage ($V_{oc}$) to collapse from 0.41 V to ~0.09 V and power density to drop by 1.5 orders of magnitude.

5. Significance

• Validation of GeSn for TPV: The study confirms that GeSn is a viable, silicon-compatible material for MWIR TPV, capable of harvesting thermal radiation without cryogenic cooling.
• Clear Optimization Path: The results indicate that the current low efficiency is not due to the bandgap or fundamental physics of GeSn, but rather material quality and device processing. The large gap between experimental and simulated performance suggests that significant improvements (orders of magnitude) are achievable through:
  • Reducing defect density (improving epitaxial growth).
  • Optimizing heterostructure design.
  • Engineering better contacts to reduce parasitic losses.
• Systemic Impact: By establishing a scalable, low-cost alternative to III-V materials, this work paves the way for the mass production of TPV systems for industrial waste heat recovery and compact power generation.
• Standardization Note: The authors highlight the critical need for standardized methods to measure Power Conversion Efficiency (PCE) in TPV systems, as current reliance on varying optical or calorimetric methods hinders accurate cross-study comparisons.

In conclusion, while current GeSn TPV devices operate far below their theoretical limits due to material defects, this work successfully establishes the foundational technology and identifies the specific engineering challenges required to unlock the full potential of silicon-compatible mid-infrared energy harvesting.