The mass production of highly crystalline carbon nanotubes (CNTs) is highly demanded, yet achieving it remains challenging due to incomplete understanding of how synthetic parameters, except temperature, affect the crystallinity of CNTs. Notably, the choice of carbon precursor significantly influences CNT synthesis, but its impact on crystallinity remains unclear. Here, we employed a data analytics approach to examine the effect of carbon precursors on CNT crystallinity during their synthesis in a fluidized bed reactor. We compared ethylene, acetylene, and a mixture of these. Using Bayesian optimization (BO), we optimized synthesis conditions to maximize IG/ID of CNTs for each precursor. Key parameters considered were reaction temperature, precursor concentration, and hydrogen concentration. We conducted three separate BO processes to evaluate the effectiveness of each carbon precursor on CNT crystallinity. The results indicated no significant difference in IG/ID of CNTs among the carbon precursors. In addition, multiple linear regression analysis did not support a synergetic effect between acetylene and ethylene. Interestingly, contour plots demonstrated consistent relationships between synthesis parameters and IG/ID across different carbon precursors. This data analytics approach allowed us to successfully assess the impact of carbon precursors on the CNT crystallinity and analyze the relationship between synthesis parameters and CNT crystallinity.

Data analytics approach to evaluating the impact of acetylene and ethylene precursors on the crystallinity of carbon nanotubes synthesized in a fluidized bed reactor
    Abstract
    1 Introduction
    2 Experimental
        2.1 Catalyst preparation
        2.2 CNT synthesis
        2.3 Bayesian optimization
        2.4 Characterization
    3 Results and discussion
        3.1 Comparison of acetylene and ethylene precursors
        3.2 Synergetic effect of ethylene and acetylene on the crystallinity of CNTs
        3.3 Characterization of as-synthesized CNTs
    4 Conclusions
    Acknowledgements 
    References

• Dong Hwan Kim(School of Chemical Engineering, Pusan National University, 2 Busandaehak‑ro 63beon‑gil, Geumjeong‑gu, Busan 46241, Republic of Korea)
• Jaegeun Lee(School of Chemical Engineering, Pusan National University, 2 Busandaehak‑ro 63beon‑gil, Geumjeong‑gu, Busan 46241, Republic of Korea, Department of Organic Material Science and Engineering, Pusan National University, 2 Busandaehak‑ro 63beon‑gil, Geumjeong‑gu, Busan 46241, Republic of Korea) Corresponding author
• Eugene Oh(LG Chem R&D Campus Daejeon, Daejeon 34122, Republic of Korea)
• Sehyun Kim(LG Chem R&D Campus Daejeon, Daejeon 34122, Republic of Korea)