Work Function Lowering of Graphite by Sequential Surface Modifications: Nitrogen and Hydrogen Plasma Treatment.

Graphite-related materials play an important role in various kinds of devices and catalysts. Controlling the properties of such materials is of great significance to widen the potential applications and improve the performance of such applications as field emission devices and catalyst for fuel cells. In particular, the work function strongly affects the performance, and thus development of methods to tune the work function widely is urgently required. Here, we achieved wide-range control of the work function of graphite by nitrogen and hydrogen plasma treatments. The time of hydrogen plasma treatment and the amount of nitrogen atoms doped beforehand could control the work function of graphite from 2.9 to 5.0 eV. The formation of a surface dipole layer and the nitrogen-derived electron donation contributed to such lowering of the work function, which is advantageous for applications in various fields.

Introduction

Two-dimensional materials consisting of sp2 carbon atoms (e.g., graphite, graphene, and graphene oxide) have been considered as promising materials for various applications such as transparent electrodes[1,2] and field emission devices.[3,4] Engineering of the structural and electronic properties of graphite sheets is essential for these applications, and then controlling the work function of these materials has a great significance. In the application for the electrodes of electronic devices, tuning the work function of graphite is a key issue because the alignment of electronic levels at the interface would dominate the device performance.[5−9] In the field emission devices, on the other hand, lowering the work function of graphite would lead to the field emitter of high efficiency. Furthermore, the graphite sheets doped with hetero-atoms show excellent catalytic activity[10,11] with the change of work function.[12]

To tune the graphite work function, various methods have been reported: doping of heteroatoms,[13,14] introduction of functional groups,[7,15,16] and so forth. The resulting change of work function was, however, small and the downward shift was especially difficult. A typical surface science method to decrease the work function of materials is usually done by depositing alkali metal atoms on the surface. In the previous reports on graphite materials, a dipole layer was formed with Cs and O atoms on the graphene surface, which decreased the work function to 3.4 eV by solution process[17] and to 1.3 eV by vacuum evaporation.[18] Although the latter method drastically decreased the work function, the surface was unstable in the atmosphere. Another way to decrease the work function of carbon-based materials is by hydrogenation of the surface. Actually, the hydrogenation of graphite decreased the work function to 4.0 eV,[19,20] and that of diamond decreased the work function to 3.5 eV, leading to a negative electron affinity surface.[21] However, the application of diamond is limited because of its low conductivity. Therefore, developing conductive materials with a low work function is required.

In the present work, we have investigated an effective surface treatment for decreasing the work function of graphite sheets by exposing a well-defined highly oriented pyrolytic graphite (HOPG) to radio-frequency (rf) hydrogen and nitrogen plasma. We found that hydrogenation decreased the work function of HOPG to 3.7 eV owing to the formation of a surface dipole layer, and nitrogen doping preceding the hydrogenation further decreased the work function to 2.9 eV. The analysis by electron spectroscopies indicated that the electron donation from nitrogen atoms assisted the decrease of the work function of graphite in addition to the formation of the surface dipole layer.

Results and Discussion

First, we investigated the effect of plasma treatment on the work function of HOPG. Figure summarizes the valence band spectra of the HOPG samples treated by various kinds of plasma treatments: N2 plasma for 30 min (N plasma), H2 plasma for 40 min (H plasma), and N2 plasma for 30 min before H2 plasma for 5 min (N–H plasma). The intensity rises steeply at the cutoff energy, from which the work function was evaluated. The cutoff energy moved upward from that of pristine HOPG (p-HOPG) after N plasma treatment, whereas it moved downward after H plasma and N–H plasma treatments. The σ peak around the binding energy of 13.7 eV[22] in p-HOPG weakened after plasma treatment, reflecting the disorder induced in the honeycomb lattice. The values of work function after each plasma treatment are shown in the figure: 4.5 eV for p-HOPG, which is well consistent with previous reports,[14,18,20] 5.0 eV for N plasma, 3.7 eV for H plasma, and 2.9 eV for N–H plasma. The work function increased after N plasma treatment, whereas it decreased after H plasma and N–H plasma treatments. We carried out several experiments to elucidate what the origin of the difference is.

Figure 1

UPS spectra of pristine and plasma-treated HOPG samples. Figures denote the work function evaluated from the cutoff energies in the spectra.

Figure shows the change of the work functions with H plasma treatment time for the p-HOPG and N-doped HOPG samples. The work function of p-HOPG decreased gradually and reached the value of 3.7 eV after 40 min treatment. This work function is much lower than that of Al (4.2 eV) which is commonly used for the electrodes in the devices of n-type organic semiconductors.[2,6,23] This decrease could be explained in terms of the surface dipole layer induced by the formation of the C–H bond.[15,21] The charge transfer from the H atom to the C atom, originating from the difference in electronegativity, would form the dipole layer on the HOPG surface. Such a kind of dipole layer was observed in hydrogen-terminated diamond, in which the dipole layer made the work function decrease by 1.5 eV at maximum.[21]

Figure 2

Change of the work function of pristine (black) and N-doped (red) HOPG samples as a function of H plasma treatment time.

To confirm the formation of the C–H bond in H plasma-treated HOPG, we conducted Raman spectroscopy and X-ray photoelectron spectroscopy (XPS). A larger Raman D peak was observed in H-HOPG than in HOPG with other kinds of treatments (Figures S1 and S2). Possible defect types in plasma-treated HOPG are vacancy, N doping, and sp3 bonding in hydrogenated carbon. Among them, sp3 bonding is considered to cause the largest D peak.[24,25] The strongest D peak in H-HOPG, therefore, suggests the formation of sp3 C–H bond by H plasma treatment.

As shown in Figure , the work function of p-HOPG reached a steady value after 40 min treatment, the reason for which could be attributed to the competing phenomena as described in the following. First, H atoms are bonded to part of the carbon atoms and the formed sp3 bond.[26] The difference in the lattice constant between the sp3 and sp2 bonds caused stress in the graphite lattice and suppressed further hydrogenation (Figure S3). Second, sputtering from continuous plasma treatment introduced vacancies in the graphite lattice,[27] which relaxed the stress and enhanced further hydrogenation.[26] Third, hydrogenated C atoms are themselves removed by the continuous H plasma treatment.[25,28] Then, hydrogenation and dehydrogenation came in equilibrium, and the work function saturated after as long as 40 min of treatment.

The change in work function for the N-doped HOPG is quite drastic as shown by the red line of Figure . The N-HOPG, which was prepared by 30 min of N2 plasma treatment, has a work function of 5.0 eV at the initial stage (Figure ). The origin of this increase can be ascribed to the charge transfer from C to N, which will be explained later. This upward-shifted work function of N-HOPG, however, decreased rapidly with H plasma treatment and reached 2.9 eV after 5 min of treatment. This value is much lower than that of hydrogen-terminated diamond and is rather close to that of Ca (2.9 eV). Further treatment gradually reincreased its work function because the doped nitrogen atoms were desorbed by H plasma treatment.

Although the N-doped HOPG itself has a larger work function than the p-HOPG, H plasma treatment changed its work function much lower than that of the p-HOPG. To elucidate this mechanism, the samples were characterized by XPS at each stage of plasma treatment. Figure a shows the N 1s spectra of N-HOPG before and after H plasma treatment. The N 1s peak consists of several components, reflecting the different chemical states. The peak was fitted with graphitic N (the binding energy of 401.4 eV), pyrrolic N (399.7 eV), and pyridinic N (398.5 eV). The sites of N atoms are schematically depicted in Figure b. According to the previous results,[13,14,29] the graphitic N donates the electron to the surrounding graphene lattice, leading to the decrease of the work function. The pyridinic N, on the other hand, increases the work function owing to the reverse electron transfer at the C–N bond.[30] Before the H plasma treatment, most of the N atoms were at the pyridinic and pyrrolic sites, which increased the work function from 4.5 eV (p-HOPG) to 5.0 eV. After the H plasma treatment, the nitrogen content decreased from 13.9 to 6.7 at. % by sputtering. It is striking that the pyridinic N completely disappeared, whereas the graphitic N and the pyrrolic N seemed to remain. Since the pyridinic N and pyrrolic N reside commonly at the edge of graphene sheets, it seems strange that only the pyridinic N was eliminated by sputtering.

Figure 3

(a) Nitrogen 1s XPS spectra of the N-HOPG sample before and after H plasma treatment. (b) Typical doping sites of nitrogen and hydrogen atoms in the graphene lattice.

We consider that the peak observed around 400 eV after the H plasma treatment originated not from the pyrrolic N but from the pyridinium N (Figure b). The pyridinium N was formed by protonation of pyridinic N, the binding energy of which was calculated to be 398.93 eV by Hou et al.[31] Nguyen and Turecek investigated the protonation of pyridine molecules.[32] They reported that the proton affinity to the N site of pyridine is 924 kJ/mol, which is larger than that to the C sites of pyridine (637–684 kJ/mol). Thus, it is plausible that the H plasma treatment sputtered part of pyridinic N and simultaneously converted the remaining pyridinic N into pyridinium N. To demonstrate this idea, we repeated the hydrogenation and dehydrogenation treatments for N-doped HOPG (Figure S4), which initially contained only graphitic N and pyridinic N components. In this experiment, H plasma clearly changes the pyridinic N to pyridinium N, and subsequent thermal annealing returned the pyridinium N to pyridinic N.

Here, we will discuss the desorption process of surface elements in graphite during plasma and thermal treatments. Desorbed elements by plasma and annealing processes are not only H atoms. Part of hydrogenated C and N atoms desorbed with H atoms in the dehydrogenation process. During the annealing of hydrogenated graphite, C atoms are desorbed with H atoms, and vacancy defects are left in the graphite surface layer (Figure S5). Similarly, the peripheral hydrogenated nitrogen such as pyridinium N seems to be also decreased by the sputtering and chemical desorption effects. Using this process, we can enhance the ratio of graphitic N/total N up to 86% (Figure S6), which is a much higher component ratio than the previous result.[13] This result shows that H plasma etching effectively removes the edge dopants and tunes the doping sites. Schiros et al. pointed out that pyridinium N slightly donates electrons to the graphene lattice in contrast to pyridinic N.[14] In spite of the very small n-doping from pyridinium N, the H plasma treatment converted the pyridinic impurities, which increase the work function, to the ones which donate electrons to the graphene lattice and causes decrease of the work function. Formation of a surface dipole layer together with electron donation in N-HOPG contributed to the larger decrease of work function than p-HOPG, for which a surface dipole layer mainly contributed to the decrease of work function. After H plasma treatment for various samples (Figure S7), the sample having the largest amount of doped nitrogen atoms showed the lowest work function since all kinds of dopants finally contribute to the decrease in the work function.

H plasma treatment to N-HOPG was found to achieve the work function as low as 2.9 eV, which is comparable to that of Ca. Generally, the surface with a low work function is unstable under atmospheric condition. To evaluate the stability, the present sample was exposed to atmosphere for 3 h. Figure a shows the change of work function at each stage of process. The initial work function of 2.9 eV moved to 3.9 eV after exposure to air for 3 h [Figure a(i,ii)]. It recovered to 3.5 eV by 1 h annealing at 130 °C in vacuum [Figure a(iii)]. It should be mentioned that the work function of N–H HOPG could be decreased as low as that of hydrogen-terminated diamond by reannealing even after exposure to air. Further annealing at 250 and 600 °C, however, reincreased the work function to 3.7 and 4.1 eV [Figure a(iv,v)]. The surface atomic composition at each stage of process was investigated by XPS, the result of which is summarized in Figure b. After the exposure in air for 3 h, the oxygen content increased to 1.6 at. % owing to the adsorption of water and oxygen molecules. The electron-withdrawing characteristics of these molecules were likely to move the Fermi level of the sample downward, and the work function of hydrogenated N-HOPG increased to 3.9 eV as shown in Figure a. Annealing at 130 °C for 1 h decreased the oxygen content to 0.9 at. %, and its work function recovered to 3.5 eV. With the annealing temperatures above 250 °C, not only the oxygen content but also the nitrogen content decreased remarkably. After annealing at 600 °C, the oxygen and nitrogen contents were decreased to 0.1 and 0.6 at. %, respectively. Elias et al. reported that hydrogen atoms desorb from hydrogenated graphene (graphane) at 350 °C in vacuum.[26] Thus, annealing above 250 °C caused this desorption of nitrogen and hydrogen atoms and increased the work function.

Figure 4

(a) Change of the work function of the N–H plasma-treated HOPG sample at each stage of process: (i) initial state, (ii) after exposure to air for 3 h, (iii) annealing in vacuum at 130 °C for 1 h, (iv) annealing at 250 °C for 1 h, and (v) annealing at 600 °C for 1 h. (b) Nitrogen and oxygen contents at each stage of process.

Finally, we would like to mention the advantages of present results. In the field of controlling the work function with hydrogenation, there are many reports about diamond materials. The hydrogen-terminated diamond decreased its work function from 5.0 to 3.9 eV and showed superior field emission characteristics. On the other hand, there are also reports that conductivity of diamond was improved by hetero-doping, and the work function of nitrogen-doped diamond was reduced to 3.1 eV by hydrogenation.[33] Even if improved by doping, the conductivity of diamond is inferior to that of graphene and graphite. By using graphite materials, we obtained a low work function material which is superior to diamond in electronic conductivity. In the view point of the surface modification method, rf plasma treatment to tune the electronic state of the graphite lattice has advantages over conventional doping methods such as chemical vapor deposition or heat treatment in the atmosphere of dopant sources.[30,34] First, the doping process could be done at lower temperatures (room temperature in the present work) under the atmosphere of easy-handling gases (nitrogen or hydrogen). Second, the doping amount of nitrogen could be tuned precisely over the wide range from subpercent to 32 at. % (Figure S7). These advantages would be helpful for the practical application of plasma to graphite-based materials.

Conclusions

In summary, we found that the combination of N-doping and hydrogenation remarkably decreases the work function of graphite. The 30 min N2 plasma and 5–10 min H2 plasma treatment achieved a work function as low as 2.9 eV, which is much lower than that of hydrogen-terminated diamond. This was realized by the formation of a surface dipole layer and the electron donation to the graphite lattice. This surface state of the modified HOPG was rather stable in ambient air, and the work function recovered to 3.5 eV even after exposure to air for 3 h. The hydrogenated graphite sheets doped with a considerable amount of nitrogen would be applied to the electrode material substituting for Ca, Al, and diamond. The effect of nitrogen doping into the graphene lattice has also attracted much attention for application to the catalyst for oxygen reduction reaction,[10] photocatalyst of g-C3N4 for water splitting and CO2 conversion,[11] lithium ion battery,[35] ultracapacitors,[36] and so forth. Thus, H plasma-induced modification of the chemical state would provide helpful information to various research fields relating to nitrogen-doped graphite. Our results demonstrated that H plasma treatment significantly changed the electronic states of graphite and provided a solid foundation for the utilization of graphite in various applications.

Experimental Section

In all experiments in this paper, we used a purchased HOPG sample: AGraphZ grade manufactured from AtomGraph AG. A p-HOPG sample was prepared by cleaving a HOPG crystal in air and annealing at 600 °C for 1 h under a vacuum of 1 × 10–6 Pa. XPS indicated no trace of impurities such as oxygen, nitrogen, and so forth. Nitrogen-doped HOPG (N-HOPG) was obtained by exposing the p-HOPG to N2 plasma under the following condition: N2 gas pressure of 1 Pa, N2 gas flow rate of 2 sccm, rf power of 10 W, and treatment time for 30 min. Hydrogen plasma treatment on the p- and N-HOPG samples was carried out under the following condition: H2 gas pressure of 10 Pa, H2 gas flow rate of 10 sccm, and rf power of 10 W. The plasma-treated sample was transferred into an analysis chamber without breaking vacuum for XPS and ultraviolet photoelectron spectroscopy (UPS) measurements.

The nitrogen concentration was the average of the graphene layers within the probing depth, six layers. In all measurements, the photoelectron emitted around the angle of 30° from the surface normal in the case of XPS and 0° in the case of UPS. The apparatus was described in the previous report.[34] A Mg Kα (hν = 1253.6 eV) light source (Thermo VG Scientific, XR3E2) and a monochromatized He I (hν = 21.2 eV) light source (SPECS, UVS 300 with TMM 302) were used for the XPS and UPS measurements, respectively. Taking account of inhomogeneity in the specimens, the work function was evaluated by the rising of cutoff energy of the UPS spectrum, which was measured with a sample bias of −5 V. The stability in atmosphere was investigated by exposing the sample to ambience for 3 h. To remove adsorbed species such as water and oxygen, the sample was annealed at 130, 250, and 600 °C before the XPS/UPS measurement. To estimate sample defects, Raman spectroscopy (JASCO, NRS-3100) was carried out with a 532 nm excitation laser.