Naslov
Predicting the transport properties of silicene nanoribbons using a neural network

Autori
Župančić, Tin ; Stresec, Ivan ; Poljak, Mirko

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Izvornik
Proceedings of Intl. Conv. MIPRO-MEET (Microelectronics, Electronics and Electronic Technology) / Skala, Karolj - Rijeka, 2020, 51-55

Skup
43rd International Convention on Information and Communication Technology, Electronics and Microelectronics (MIPRO 2020) ; Microelectronics, Electronics and Electronic Technology (MEET 2020)

Mjesto i datum
Opatija, Hrvatska, 28.09.2020. - 02.10.2020

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
quantum transport ; silicene nanoribbon ; mobility ; bandgap ; neural network ; TensorFlow ; Keras

Sažetak
Atomistic quantum transport simulations are used to generate the electronic and transport properties of 10, 000 realistic silicene nanoribbons (SiNRs) with edge defects. This ensemble of 20 nm-long and 2.1 nm-wide SiNRs is divided into the training and inference set for the artificial neural network (ANN) employed for the prediction of edge-defect-limited carrier mobility from the known values of bandgap and nanoribbon conductance. We find that an optimized ANN with 3 hidden layers can predict SiNR mobility values and variability histograms with acceptable accuracy, thus providing a useful supplement to atomistic quantum transport simulations that take several hours or days for large device ensemble sizes.

Izvorni jezik
Engleski

Znanstvena područja
Elektrotehnika

POVEZANOST RADA