MD Simulations of Hydrogen Plasma Interaction with Graphene Surfaces David B. Graves Chaire d’excellence, Fondation Nanoscience, Grenoble, France Chemical & Biomololecular Engineering, U.C. Berkeley August 5-9, 2013 21rst ISPC Cairns, Australia

Acknowledgements Emilie Despiau-Pujo, CNRS/UJF-Grenoble1/CEA LTM Alexandra Davydova, CNRS/UJF-Grenoble1/CEA LTM Gilles Cunge, CNRS/UJF-Grenoble1/CEA LTM Laurence Maguad, CNRS/UJF-GRENOBLE 1 INSTITUT NÉEL Laure Delfour, CNRS/UJF-GRENOBLE 1 INSTITUT NÉEL Dave Humbird, former UCB PhD (US DoE) Joe Végh, former UCB PhD/postdoc (Lam Research Corporation) Ning Ning, former UCB postdoc (Paris) Fondation Nanoscience: Chaire d’excellence (Grenoble, France) Laboratoire des technologies de le Microélectronique (LTM/CEA) Lam Research/Hewlett Foundation: UCB Distinguished Chair

Motivation : Graphene may have uniquely promising properties §  1-atom-thick planar sheets of sp2-bonded carbon atoms packed in a honeycomb lattice §  2D structure + outstanding physical, chemical, thermal, optical and mechanical properties

Promising candidate for many novel applications… Transparent electrodes, flexible touch-screens

Hydrogen storage

High-speed transistors

Aviation materials

… but relies our capability to grow and integrate it into sophisticated devices

What are Plasma Processing Implications for Graphene? •  Single or few-layer graphene is certainly vulnerable to ion-induced damage •  Graphene often processed via remote (downstream) plasma so only radicals impact surface •  But recent results suggest that graphene can be processed with conventional plasma •  Furthermore, pulsed plasma technology offers real promise to exploit benefits of plasma with no damage •  However, precise control of ion energies and radical fluxes will be crucially important: atomistic simulation

Graphene can be used as FET channel material if bandgap can be created

Large area

GNR

Bilayer

Nanoribbon width must be carefully controlled for band gap and a properly functioning transistor (on/off ratio)

Molecular Dynamics: Basic principles Simulate the motion of interacting particles through integration of Newton equations of motion, using potentials that can account for bond breaking & formation.

fi

- compute trajectories (ri,pi) of N interacting atoms as function of time Principles * classical description of atomic motion * semi-empirical description of the e- structure based on many-body potentials for each atom i in the system : ..

f i = m i r i = −( U(r1 , r2 ,...) = ∑ U 2 (rij ) + i< j

* statistical mechanics

∂U ) ∂ri

∑U

i < j< k

3

(ri , rj , rk ) + ...

MD: Impact simulation as a model of bombardment for H2 plasma Top surface exposed to ion and neutral flux (random location)

H

H+

Geometry Surface Area 400 - 900 A2 Thickness 5 - 30 A Number of atoms 160 - 1000 Boundary Conditions Periodic in x & y directions (except for GNR) 1 single atom fixed on each layer Charge Ions = fast neutrals (Auger neutralization) Injection Chemical nature inert or reactive Energy [0.026 ; 100] eV Incident Angle [0° ; 90°] Dose (fluence) [1016 ;1017] impact/cm2

H – graphene infinite layer or nanoribbon interactions z

4A y

x

free ZZ-edges

free AC-edges

§  Single-layer graphene surfaces semi-infinite or GNR with free edges temperature : 0K §  H(+) impacts normal incidence specific locations injection height : 4A above surface low energies : [0.026 - 25] eV

C-H bond formation (0.8eV H+ - 25ps relaxation) § H chemisorption - outward motion of C atom in C-H bond

1.5 2

Potential energy (eV)

1.0

A

B

C

C-sp

2

§  dC-H=1.11A / Echem=0.66 eV

0 C-sp

-2

3

0.5 -4 0

0.0

1

2

3

A B

-0.5

C

§  in agreement with DFT studies*, UPS, STM and HREELS experiments**

-1.0 -1.5 0.5

1.0

1.5

§  activation barrier – energy necessary to distort lattice -> repulsive force due to delocalized πe- in sp2 hybrid-state

2.0

2.5

3.0

3.5

distance of H above the surface (A)

chemical binding of H -> local sp2/sp3 rehybridization -> structural modifications of graphene sample *Boukhvalov PRB 77 (2008) /Šljivančanin JCP 131 (2009) **Ruffieux PRL 84 (2000) / Zecho JCP 117 (2002)

Influence of H incident energy & surface temperature Statistical runs Tcell : 0-300-600 K / Ei : [0.1-200] eV

H+ H+

0.8

reaction rate

Each {Ei, Ti} : 200 impacts at normal incidence and random location on a refreshed cell

1.0 0K 300K 600K

reflection

penetration

0.6

0.4

H+ H+

H+

0.2

0.0

0.1

absorption

1

10

Incident H energy (eV)

Main effect of temperature at low energies -> threshold/energy barrier for H adsorption decreases with increasing T

100

Cumulative bombardment by thermal radicals (300K): Exp

0.8

reaction rate

Behr et al, JVSTB 28 (2010)

1.0 0K 300K 600K

reflection

penetration

0.6

0.4

0.2

0.0

0.1

absorption

1

10

100

Incident H energy (eV)

Yang et al, Adv.Mater. 2010 (22)

Xie et al, J. AM. CHEM. SOC. 2010 (132)

MD results in agreement with TEM/AFM analyses which show that H etching occurs preferentially at graphene edges or defect sites

MD Simulation of Edge Etch: Pre-Hydrogenated Edges Speed Computational Time

C Etch vs. H Fluence: Initial Stages Note: C-C Bond Breaking

Edge C Start to Etch

Nearly All Edge C Have Etched: Selective Edge Etch Observed

Note Edge C Only Etch: Unifiorm Radial Etching initial

final

C Etches from Edge as C or C2 Dimers

Note lack of H in etch products!

Summary and Concluding Remarks 1.  MD simulations appear to be in qualitative agreement with experiments that show edge-selective etching. 2. Initial hydrogenation of zig-zag edge leads to subsequent hydrogenation in ‘second row’ C; this introduces stress and causes a ‘crimp’ to form at edge of GNR. 3. Remarkably, all etch products are either C or C2: no H in etch products. 4. H impact does affect C loss; understanding mechanisms is in progress.

I

Opening of C-C bond in the hexagone

Carbon sputtering from the strain

Two carbon atoms were left from sides

II

Single carbon left from the edge hexagone

 Three carbon atoms left the strain as C2 and C

 Two free C-C bonds on the edge reconstructed into pen

III

C-C bond in the strain stressed and broken

One by one C atoms leave the chain

IV

Single hexagone on the edge lost two carbon atoms as C2 etching product

We assume that impacting hydrogens brings additional vibration into the system Thus, thermal fluctuation between atoms cause the C sputtering from the edges There is no CxHy formation due to absence of dangling bonds in triple-oriented carbons All these results are under investigation.