[Paper Review] Enhancing thermoelectric properties of isotope graphene nanoribbons via machine learning guided manipulation of disordered antidots and interfaces

Xiang Huang, Shengluo Ma+5 more|arXiv (Cornell University)|Jul 12, 2022

Advanced Thermoelectric Materials and Devices4 citations

TL;DR

This study enhances thermoelectric performance in isotope graphene nanoribbons (AGNRs) by using machine learning to optimize disordered antidots and aperiodic isotope superlattices. By combining Green's function simulations with ML-guided design, the authors achieve a ZT of 0.894—5.69× higher than pristine AGNRs—through continuous isotope interfaces that suppress phonon transport more effectively than isotope substitution alone.

ABSTRACT

Structural manipulation at the nanoscale breaks the intrinsic correlations among different energy carrier transport properties, achieving high thermoelectric performance. However, the coupled multifunctional (phonon and electron) transport in the design of nanomaterials makes the optimization of thermoelectric properties challenging. Machine learning brings convenience to the design of nanostructures with large degree of freedom. Herein, we conducted comprehensive thermoelectric optimization of isotopic armchair graphene nanoribbons (AGNRs) with antidots and interfaces by combining Green's function approach with machine learning algorithms. The optimal AGNR with ZT of 0.894 by manipulating antidots was obtained at the interfaces of the aperiodic isotope superlattices, which is 5.69 times larger than that of the pristine structure. The proposed optimal structure via machine learning provides physical insights that the carbon-13 atoms tend to form a continuous interface barrier perpendicular to the carrier transport direction to suppress the propagation of phonons through isotope AGNRs. The antidot effect is more effective than isotope substitution in improving the thermoelectric properties of AGNRs. The proposed approach coupling energy carrier transport property analysis with machine learning algorithms offers highly efficient guidance on enhancing the thermoelectric properties of low-dimensional nanomaterials, as well as to explore and gain non-intuitive physical insights.

Motivation & Objective

• To overcome the challenge of coupled electron and phonon transport in nanoscale thermoelectric materials.
• To explore how structural disorder and isotope engineering affect thermoelectric performance in graphene nanoribbons.
• To develop a machine learning-guided design framework for optimizing multifunctional transport in low-dimensional nanomaterials.
• To identify non-intuitive physical mechanisms that enhance thermoelectric efficiency in isotope-modified AGNRs.

Proposed method

• Combines non-equilibrium Green's function (NEGF) formalism with atomistic simulations to compute electron and phonon transport in AGNRs.
• Introduces disordered antidots and aperiodic isotope superlattices as structural tuning parameters.
• Employs machine learning algorithms to navigate the high-dimensional design space of antidot placement and isotope distribution.
• Uses ZT (figure of merit) as the optimization target, with thermoelectric performance evaluated via electrical conductance and thermal conductance.
• Trains ML models on simulated data to predict optimal configurations with high ZT values.
• Validated results through physical analysis to reveal that continuous isotope interfaces perpendicular to transport direction are key to phonon scattering.

Experimental results

Research questions

• RQ1 How does the introduction of disordered antidots affect the thermoelectric performance of isotope-modified graphene nanoribbons?
• RQ2 What role do aperiodic isotope superlattices play in suppressing phonon transport and enhancing ZT?
• RQ3 Can machine learning effectively navigate the complex design space of nanostructured AGNRs to identify high-performance configurations?
• RQ4 Why is the antidot effect more effective than isotope substitution in improving thermoelectric properties?
• RQ5 What physical mechanisms underlie the enhanced ZT in the optimal ML-designed structure?

Key findings

• The optimal AGNR structure achieved a ZT of 0.894, representing a 5.69-fold improvement over the pristine graphene nanoribbon.
• The highest ZT was achieved at interfaces of aperiodic isotope superlattices with disordered antidots, indicating synergistic effects.
• Carbon-13 atoms preferentially form continuous, transverse barriers that suppress phonon propagation, significantly reducing thermal conductivity.
• The antidot effect outperformed isotope substitution in enhancing thermoelectric performance, highlighting its superior scattering efficiency.
• Machine learning identified non-intuitive structural configurations that maximize ZT, revealing physical insights beyond conventional design principles.
• The coupling of NEGF simulations with ML provided an efficient, physics-informed optimization pathway for low-dimensional thermoelectric materials.