Machine Learning-Aided Band Gap Engineering of BaZrS₃ Chalcogenide Perovskite

Area of Science:

• Materials Science
• Solid State Chemistry
• Photovoltaics

Background:

• Barium zirconate sulfide (BaZrS₃) is a promising non-toxic chalcogenide perovskite for optoelectronic applications.
• It exhibits a direct band gap (1.7-1.8 eV), high absorption, and good carrier mobility, suitable for tandem solar cells.
• However, its band gap is larger than the optimal 1.3 eV for single-junction solar cells, necessitating band gap reduction through doping.

Purpose of the Study:

• To identify and predict optimal dopants for BaZrS₃ to achieve a band gap within the Shockley-Queisser limit for single-junction solar cells.
• To experimentally validate the effectiveness of calcium (Ca) doping at the barium (Ba) site in BaZrS₃ for photovoltaic applications.

Main Methods:

• Utilized first-principles calculations and machine learning algorithms to screen potential dopants for BaZrS₃.
• Synthesized Ca-doped BaZrS₃ (Ba_1-xCa_xZrS₃) and compared its properties with previously reported Ti-doped BaZrS₃ (Ba(Zr_1-xTi_x)S₃).
• Measured and compared the photoluminescence properties and band gaps of the synthesized doped perovskites.

Main Results:

• First-principles calculations predicted Ca at the Ba site or Ti at the Zr site as optimal dopants.
• Synthesized Ba_1-xCa_xZrS₃ with <2 atom % Ca doping demonstrated a significant band gap reduction from ~1.75 eV to ~1.26 eV.
• Ca doping at the Ba site proved superior to Ti doping at the Zr site for achieving the desired band gap tuning for photovoltaics.

Conclusions:

• Partial Ca doping at the Ba site is an effective strategy to tune the band gap of BaZrS₃ for high-efficiency single-junction solar cells.
• The achieved band gap of ~1.26 eV with Ca doping is within the optimal range for photovoltaic applications.
• Ca-doping presents a more advantageous approach than Ti-doping for developing next-generation photovoltaic materials based on BaZrS₃.