Work function of the oxygen functionalized graphenic surfaces - Integral experimental and theoretical approach

The effect of oxygen plasma functionalization of graphenic surface was thoroughly investigated experimentally (SEM, XPS, Raman spectroscopy, work function measurements, TG) and corroborated by DFT molecular modeling using periodic slabs (work function, surface dipoles, E f /DOS). It was found that the introduction of oxygen substantially modified the electrodonor properties of the graphenic surface and the work function changed in a non-monotonous way upon the plasma treatment time. At the short contact time (5-30 s) the work function significantly increased from 4.2 eV (parent graphenic surface) to 5.6 eV and after passing the maximum value it reached plateau at the level of 5.2 eV after 100-150 s. The XPS revealed that the plasma treatment leads to an increase in the oxygen surface concentration (up to ~10 at.%). A comparison of SEM images and Raman spectra of unmodified and plasma-modified graphenic surfaces indicated that the induced changes are limited to the outermost surface. Indeed, the TGA stability profiles did not indicate any bulk structural changes. Based on the experimental and DFT results, the molecular models of the surface modification process was proposed taking into account the various location of surface oxygen functionalities. At the early stage of the plasma treatment, a generation of strongly polar surface functional groups (surf-OH with the local dipole moment of 2.7 D) leads to the formation of an electrostatic potential barrier for electron transfer from the surface (observed increase in work function). Prolonging modification results in the insertion of oxygen heteroatoms into the carbon surface, more uniform electron density distribution, and hence work function decrease. Since the electronic properties of graphenic materials play a key role in their various applications, the obtained results provide rational guidelines for their design and optimization.

. It is worth mentioning that to preserve the superior native bulk properties o f car bon materials, plasma modification seems to be more favorable.
The oxygen plasma is considered as an efficient method to tailor the surface oxygen functional groups by reaction o f gas-phase oxygen reactive species with surface carbon atoms [18,19]. The usage of plasma permits carbon surfaces (e.g., fibers, nanotubes, graphene-based mate rials) to be functionalized chemically by the generation o f new oxygencontaining surface functionalities to improve their wettability and/or electrocatalytic activity [20][21][22][23].
The application o f plasma allows to limit the modification to the outermost surface and also provides the way to carry out the modifi cation in a precisely controlled way. This can be achieved by adjusting the working parameters such as generator power, oxygen partial pres sure, and time o f exposure. Additionally, the method is environmentally friendly, efficient in terms o f energy and cost, and fast (requires a short time o f seconds/minutes) [18,24,25].
The introduction o f oxygen moieties to carbon material surfaces changes dramatically their wettability, acid-base properties, and leads to the formation o f new active centers for adsorption [19,26]. It was also shown that by controlling the surface oxygen concentration and its chemical nature, the electronic properties o f carbon materials can be precisely tuned. In this context, the work function was found to be a sensible and convenient parameter for quantification o f the extent of functionalization o f carbon surfaces, such as graphite, graphenic, and nanotubes [19,23,24]. Besides the extensive experimental work on carbon material synthesis, characterization and applications, the mo lecular level description is still lacking, in particular when the surface functionalization is addressed.
The computational studies, in both cluster and periodic boundary conditions (PBC), serve as a complementary technique to the experi mental ones. The cluster approach is easier applicable for the studies of the local properties and the determination o f the entropic contributions for the interactions o f the solid surface with the gas phase. In turn, the PBC reproduces the solid-state properties directly stemming from the electronic structure (band structure, electrostatic potential, etc.) and also presents generally a more unbiased way o f the computational model design. The rich literature on this topic is available, to cite only a few review articles [27,28].
This study aimed to investigate the functionalization o f graphenic material by surface oxygen species generated via plasma treatment. The applied approach takes advantage o f combining experimental charac terization o f the functionalized graphenic surfaces (Scanning Electron Microscopy, X-ray Photoelectron Spectroscopy, Raman Spectroscopy, Thermogravimetry and work function measurements) and molecular interpretation based on the periodic DFT modeling. The surface modi fication is discussed in terms of the formation of surface dipoles, defects and changes in work function and perturbation o f density o f states of the graphenic material.

Samples preparation and modification
The conductive graphenic sheets (25 |im thickness, 2 g-cm~3) sup plied by Graphene Laboratories Inc. were used in the study. The samples were cut into 1 x 1 cm coupons and cleaned in isopropanol. The surfaces o f graphenic samples were modified with oxygen plasma treatment   Table 1.

Samples characterization
Directly after the modification, samples were investigated with the methods described in detail below.

Scanning Electron Microscopy (SEM)
To evaluate the surfaces after plasma treatment the images o f graphenic sheets were taken with the use o f a Hitachi S-4700 scanning electron microscope. The following settings were applied: operating voltage 10 kV, working distance 11.1 mm.

X-ray Photoelectron Spectroscopy (XPS)
The surface composition was investigated with the use o f XPS in an ultra-high vacuum system, equipped with an SES R4000 (Gammadata Scienta) analyzer. The monochromatic Al Ka source (1486.6 eV) oper ated at 350 W was applied. The spectra were obtained at a takeoff angle o f 90°. The vacuum in the spectrometer chambers was better than 5 x 10~9 mbar. Due to the static charging o f the samples, correction o f the binding energy (BE) scale was adjusted by setting the C 1s peak at 285 eV. The acquired spectra were analyzed using Casa-XPS 2.3.15 software.
The integration o f narrow O1s and C1s spectra were used for evaluation o f surface oxygen content introduced via plasma treatment.

Work Function measurements
The work function values (O ) o f the investigated graphene samples were determined based on measurements o f the contact potential dif ference (VCPD) carried out by the Kelvin method with a KP6500 probe (McAllister Technical Services), as described in detail in [23,24]. The work function values were calculated as an average o f 60 independent measurements for each sample based on the relation: Osample = Oref -eVCPD. The measurements were performed at ambient conditions (at mospheric pressure, room temperature).

Quantum-chemical calculations
To study the influence o f the plasma treatment on the graphenic surface at the molecular level, quantum-chemical modeling was applied,

Slab model
A sequence o f models, with protruding -OH groups and with point defects occupied by O atoms, was designed to model the initial and late phases of the plasma treatment, respectively. The models were designed based on the experimental indications concerning the chemical composition and the superficial character of the plasma treatment in the applied conditions. Concerning the density o f oxide moieties on the surface, the Lerf-Klinowski model o f graphene oxide was taken as a basis [17,41]. In the reference model, the occupation density was up to nine entities per 100 A 2. Taking into account the mild plasma treatment conditions and the model surface area, in this article, the functional group density was used up to five entities per 84 A 2.
The general description o f the construction o f the graphenic bilayer was included in our previous article [17]

Results and discussion
There are several reports on plasma functionalization o f various carbon materials showing that upon controlled treatment, the process can be limited to the thin surface layer [17,43,44]. This is o f particular the maximum value o f work function at 5.3 eV was found, whereas for the power o f at least 50 W the maximum is shifted to 5.6 eV and seems to be comparable within experimental error. The data is consistent with the XPS results, confirming the increased surface oxygen content for a higher power. Based on the data collected in Fig. 4, it can be also noted that for graphenic surface modification by oxygen plasma for higher plasma generator power (> 5 0 W), the work function maximum is reached after shorter modification times. Since the obtained results are in line with the previously reported work function changes upon plasma treatment for graphite (flat surface) [23] and multiwalled carbon nanotubes (curved surfaces) [19] it can be deduced that the observed

Computed work function vs surface functionalization
The examples o f charge density isosurfaces tinted with the electro static potential for unmodified and modified (defected) by the oxygen graphenic surfaces are presented in The work function changes can be further rationalized by analyzing the dipole moments (Table 2), which influence the "vacuum level" (the plateau/asymptotic value between the slab and its translational image above the vacuum layer) o f the averaged electrostatic potential and the DOS ( Fig. 6, Fig. 7). In the vicinity o f the Fermi level, the density of electronic states is only slightly perturbed while the normal component  as reported in our previous work [2 3 ]. Therefore, we focus here on this particular surface species while analyzing the oxygen surface coverage.
Since the native graphene model bilayer does not exhibit a dipole  Fig. 8 together with the differential density plots for the calculated models.
From the way the dipole moments have been derived, they are the components normal to the surface ( "Z" components) and they can be The harsher surface functionalization leads to breaking of the surface carbon bonds and formation o f the surface vacancies, occupied by the oxygen atoms ( Fig. 5). Here, the surface dipole moment is much lower, which is in-line with the almost flat geometry of such defects, compared to the protruding -OH ad-groups. Due to the stronger modification of the electronic structure o f the C atoms adjacent to the vacancy, the DOS close to the Fermi level is more changed (Fig. 7). Namely, for the single and double vacancy, the feature is shifted toward the negative energy by a few tenths o f the eV which is reflected by the lowering o f the work function compared to the pristine graphene bilayer (see Table 3). Since the formation o f -COOH groups may also contribute to changes in the graphenic DOS structure, they were also simulated. As the overall pic ture is dominated by -OH and probability o f -COOH formation iss low, the DOS structure for the latter one is presented in Supplementary In formation (