Direct fluorination of graphene: A theoretical and computational study of its formation and of the resulting magnetic and electronic properties

Open Access
- Author:
- Aditya, Piali Mitil
- Graduate Program:
- Physics
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- November 24, 2015
- Committee Members:
- Jorge Osvaldo Sofo, Dissertation Advisor/Co-Advisor
Jorge Osvaldo Sofo, Committee Chair/Co-Chair
Kristen Ann Fichthorn, Committee Member
Renee Denise Diehl, Committee Member
Mauricio Terrones Maldonado, Committee Member - Keywords:
- graphene
fluorine
DFT
spin
magnetic moment
adatom
adsorption - Abstract:
- The adsorption of fluorine changes the electronic, mechanical, and magnetic properties of graphene. While graphene is an excellent conductor and a semimetal, fully fluorinated graphene is an insulating wide bandgap semiconductor. The electronic properties of graphene can be modified by controlling the adsorbate concentration to produce conducting, semiconducting or insulating components for nanoscale electronic devices. The high electronegativity of fluorine makes it very reactive to the graphene sheet resulting in structures that are stable under ambient conditions. Moreover,recent reports of spin 1/2 paramagnetism in graphene has invigorated research efforts in this field due the possibility of spin transport devices. While there is a lot of speculation about the origin of the spin, no clear theoretical explanation exists in the literature. Semi local DFT functionals predict that the fluorine adatom is non-magnetic, whereas calculations with hybrid functionals indicate a local moment of 1μB. However, neither approaches can explain the trends in the experimentally observed spin concentration as a function of fluorination percentage. After an introduction in Chapter 1 and an overview of our methods in Chapter2, in Chapter 3, using density functional theory (DFT) we show that in highly fluorinated graphene, small regions of unfluorinated carbon atoms produce localized magnetic states at the fermi-level. We study the shape and size dependence of these regions on the net spin and find that most odd clusters have a net spin of 1/2 while most even clusters have zero spin. We construct a minimal tight binding model that captures the low energy response of DFT and describes the localized magnetic states produced by the unfluorinated carbon atoms. This model is then solved exactly to include the e!ect of excited states in the magnetic response and go beyond the mean field predictions of DFT. The model for magnetic carbon regions, when combined with large scale molecular dynamics methods that simulate the surface configurations formed upon fluorination, can provide a theoretical description of the magnetic response of partially fluorinated samples. However, commonly used semilocal functionals do not provide a physically correct description of the interaction of dilute fluorine adsorption on graphene. These functionals predict a fractional charge on fluorine at a distance of 10Å from the surface (delocalization error), however at such large distances the fluorine atom should interact with the graphene sheet through weak Van Der Waals forces. Classical potentials that are fit to the incorrect behavior of DFT inherit these errors and produce inaccurate fluorine interactions in dynamical processes. We describe the limitations of these functionals in chapter 4 and provide a physical model that predicts a physically reasonable fluorine interaction behavior as a function of distance from the surface. In chapter 5, we turn towards a reactive force field method, ReaxFF to simulate the fluorination of graphene. The screening e!ect of fluorine atoms already adsorbed on the surface causes the charge transfer to fluorine atom to be negligible beyond 5Å from the surface. This is important because the fluorine atoms will not be accelerated over large distances towards the surface. We study the adsorption process and surface configurations formed upon fluorination and find that though initially the attachment starts in a correlated fashion, we do not see the formation of large fluorographene (CF) like domains. The fluorine atoms adsorb at multiple sites over the lattice, with not enough time to diffuse and form domains. This creates several small unfluorinated carbon regions and applying the model developed in Chapter 3 overestimates the magnetic response. In conclusion, the magnetic model developed in Chapter 3 is suitable for describing the magnetic response in highly fluorinated graphene with well defined domains of CF like regions as described in Appendix A. For dynamical simulations that produce non-equilibrated configurations, the unfluorinated carbon atoms are not well localized and their interactions are likely to produce a different spin response. This being said, the classical potentials used to simulate the atomic configuration upon fluorination are dependent on DFT calculations of physical parameters like binding energy, charge, bond lengths, etc. Thus, a self interaction error free exchange correlation functional would provide an updated description of the dynamics and interaction on surfaces.