Atomic Vacancy Engineering of Graphitic Surfaces: Controlling the

May 13, 2009 - The selective introduction of atomic-scale defects with tunable density in graphite, graphene, and carbon nanotubes is highly desirable...
0 downloads 9 Views 2MB Size
J. Phys. Chem. C 2009, 113, 10249–10255

10249

Atomic Vacancy Engineering of Graphitic Surfaces: Controlling the Generation and Harnessing the Migration of the Single Vacancy J. I. Paredes,* P. Solı´s-Ferna´ndez, A. Martı´nez-Alonso, and J. M. D. Tasco´n Instituto Nacional del Carbo´n, CSIC, Apartado 73, 33080 OViedo, Spain ReceiVed: February 20, 2009; ReVised Manuscript ReceiVed: April 8, 2009

The selective introduction of atomic-scale defects with tunable density in graphite, graphene, and carbon nanotubes is highly desirable, as it could afford a better control of their properties, but difficult to realize in practice. Here, we present a plasma-based chemical approach for the selective generation of single vacancies on the graphite surface with densities between ∼5 × 102 and 3 × 105 µm-2 and apply it to the investigation of the migration behavior of this type of defect. Through scanning tunneling microscopy (STM) observation of vacancy-decorated graphite surfaces heat-treated to several different temperatures, the migration barrier for the single vacancy was deduced to be ∼0.9-1.0 eV but could be reduced to ∼0.7 eV via interaction with the STM tip. This constitutes the first direct experimental measurement of the migration barrier for the single vacancy in graphite that is consistent with theoretical calculations. 1. Introduction The presence and evolution of atomic-scale defects in graphite and related carbon nanostructures, such as graphene or carbon nanotubes, play a prominent role in their physical and chemical properties.1-5 Thus, the intentional introduction of defects, most notably by electron or ion irradiation, constitutes a powerful tool for the engineering of these materials.2,6,7 One of the most important types of defects are atomic vacancies, and in this regard many theoretical and experimental studies have addressed their strong influence on the characteristics of carbon materials with graphitic structure. For example, atomic vacancies have been shown to control the conductance8,9 and mechanical properties10 of carbon nanotubes and are also thought to allow the coalescence between individual tubes in carbon nanotube bundles.11 Likewise, vacancies have been proposed as catalytic sites in graphene and carbon nanotubes for the thermal dissociation of water to produce hydrogen12 and as the origin of ferromagnetism observed in irradiated graphite samples.13,14 The generation of atomic-scale defects on carbon structures by the usual “physical” approach, based on irradiation with electrons or ions, is generally a nonselective process, whereby several different types of defects are introduced. These include single vacancies, multivacancies, interstitials, bound interstitialvacancy pairs, nonhexagonal carbon rings, or even more complicated defect structures.2,15-17 The production of a defect population of monodisperse type (i.e., consisting exclusively of one type of defect) and controllable density would be greatly beneficial, as it could provide better control of the properties of the material, which can be crucial for many applications, but this has so far not been realized. Here, we take the first steps in such a direction and report the use of a “chemical” approach (mild oxygen plasma treatment) to generate a monodisperse population of single vacancies with tunable density on a graphite surface. Furthermore, as a first direct application of this possibility, we tackle a long unresolved issue in fundamental carbon science: the migration of the single vacancy along a graphene sheet, which has * Corresponding author. Telephone: (+34) 985 11 90 90. Fax: (+34) 985 29 76 62. E-mail address: [email protected].

implications, e.g., in the evolution of irradiated graphite and carbon nanostructures.1,2 Early reports based on indirect experiments suggested that the single vacancy in graphite migrates along the graphene plane with an activation barrier of 3.1 eV.18 However, this result has been seriously challenged by recent theoretical calculations, which predict much lower migration barriers for the single vacancy1,19-21 and attribute the large experimental value to more complex defects.1,19 Such argument is reasonable, as the polydispersity in defect type generated on electron- or ion-irradiated graphite implies that assignment of activation energies derived from indirect experiments to specific mechanisms (for instance, to the migration of single vacancies) is far from straightforward. Nevertheless, to date no direct experiment has been carried out to validate one of the two possibilities (indirect experiment vs theory) or even to rule out both. Furthermore, there are significant discrepancies in the theoretical predictions of the migration barrier, ranging from a value of 1.7 eV,1,19 which is now the most widely accepted, to as low a value as ∼1 eV.20 With the selective generation of single vacancies reported in this work, it becomes possible to gain valuable insight into the migration of this type of defect by studying their aggregation behavior at different temperatures with scanning tunneling microscopy (STM). 2. Experimental Section Grade ZYH highly oriented pyrolytic graphite (HOPG) samples, acquired from Advanced Ceramics Corp. (Cleveland, OH), were used for the present study. Before any manipulation, the graphite specimens were cleaved in air to obtain fresh, pristine basal plane surfaces. Single vacancies were produced on such HOPG surfaces by means of mild oxygen plasma treatments. The treatments were carried out in a Technics Plasma 200-G apparatus (Kirchheim bei Mu¨nchen, Germany), which employs 2.45 GHz microwave (MW) radiation to activate the plasma. The MW radiation was generated by a magnetron and transferred via a waveguide to the treatment chamber (quartz reactor), where the plasma was created and the HOPG samples were placed. O2 (99.999% pure) was used as the plasma gas at a working pressure of 1.0 mbar. To create only isolated single vacancies and avoid the formation of extended defects (multi-

10.1021/jp901578c CCC: $40.75  2009 American Chemical Society Published on Web 05/13/2009

10250

J. Phys. Chem. C, Vol. 113, No. 23, 2009

Paredes et al.

Figure 1. (a-d) Representative nanometer-scale STM images of single vacancy-decorated graphite surfaces prepared by MW oxygen plasma treatment using different combinations of MW power/treatment time: (a) 30 W/4.5 s, vacancy density ∼3 × 103 µm-2; (b) 30 W/6 s, vacancy density ∼2.5 × 105 µm-2; (c) 80 W/4 s, vacancy density ∼6 × 104 µm-2; (d) 150 W/4 s, vacancy density ∼3 × 105 µm-2. (e) Typical atomic-scale STM image of a single vacancy-decorated graphite surface prepared by MW oxygen plasma treatment at 80 W for 4 s. (f) Lateral size distribution for the single vacancies as deduced from the STM images. The histogram ordinate represents normalized frequency.

vacancies and nanometer-sized pits), very short treatment times (typically restricted to a few seconds) were employed. The vacancy density on the HOPG surface could be tuned through proper combination of the MW power used to activate the plasma and the treatment time. As a general rule, increasing the MW power or the treatment time led to higher densities of single vacancies. For the migration experiments, typical treatments were carried out at 80 W for 4 s, as this combination yielded a suitable density of vacancies (∼6 × 104 µm-2). The migration of these defects was investigated by observing their aggregation behavior following heat treatment of the single vacancy-decorated HOPG surfaces at different temperatures. The heat treatments were carried out in a horizontal quartz reactor (20 mm of internal diameter) under a flowing (80 mL/min) Ar atmosphere. A linear heating rate of 2 °C/min was employed. The HOPG samples were kept at the desired heat treatment temperature for 60 min and then allowed to cool down to room temperature. The temperature was measured with a Pt/Rh thermocouple placed inside the reactor and close to the HOPG sample. Scanning tunneling microscopy (STM) was employed to visualize the plasma-generated defects. The STM measurements were accomplished under ambient conditions in a Nanoscope IIIa Multimode apparatus, from Veeco Instruments (Santa Barbara, CA). Imaging was performed in the constant current mode (variable height) using mechanically prepared Pt/ Ir (80/20) tips and with typical tunneling parameters of 100 mV (bias voltage) and 0.5-2.0 nA (tunneling current), unless otherwise stated. The quality and stability of the tips were verified by first scanning the surface of pristine HOPG. Only those tips that consistently yielded high quality atomic resolution images were used for the vacancy migration studies. Several different tips were employed to check for the reproducibility of the results. To make sure that the changes observed in the STM images of heat-treated HOPG were effectively caused by the heat treatment, the starting vacancy-decorated samples were prepared by MW plasma in batches of two pieces. After plasma treatment, both pieces were examined by STM to confirm the generation of isolated single vacancies with the expected density. Then, one of the pieces was subjected to heat treatment, whereas the other was kept as a control sample. Following heat treatment, both samples were examined again by STM and compared.

Additional microscopic characterization was carried out by means of conducting probe atomic force microscopy (CP-AFM). CP-AFM was accomplished with a Nanoscope V Multimode system equipped with a TUNA extension. Pt/Ir-coated rectangular Si cantilevers with a nominal spring constant of 0.2 N m-1 were employed. Measurements were performed under ambient conditions with typical applied loads of several tens of nanonewtons and bias voltages of several hundred millivolts. 3. Results and Discussion Gas-phase oxidation, e.g., in molecular or atomic oxygen,22,23 ozone,24 or MW oxygen plasma,25,26 is known to lead to the removal of carbon atoms from the surface of graphitic materials (etching), thus constituting a potentially useful approach for the controlled generation of atomic vacancies on graphite. In the case of MW oxygen plasma, and in contrast to other types of plasmas, etching proceeds only on the basis of chemical reactions between the active species from the plasma (activated oxygen species) and the carbon atoms, leading to the removal of the latter from the graphite basal surface in the form of CO or CO2 molecules (chemical attack). Physical attack, that is to say, the creation of defects by bombardment of energetic particles, is not present in this type of plasma.25 More importantly, the chemical attack in the MW oxygen plasma can be controlled to such an extent that it is possible to prepare graphite samples which have experienced only the very first stage of etching, i.e., the abstraction of individual carbon atoms from the basal planes, giving rise to the creation of single vacancies on the surface. Furthermore, it was observed that this method allows tuning of the vacancy density in a broad range, approximately between 5 × 102 and 3 × 105 µm-2. This could be accomplished by adjusting both the MW plasma power and the treatment time within certain limits. As a general rule, to generate only single vacancies and avoid their expansion into multivacancies and nanometer-sized pits, very short treatment times (restricted to a few seconds) had to be employed. For a given treatment time, higher vacancy densities were obtained by increasing the MW power. Similarly, increasing the plasma treatment time resulted in higher densities of single vacancies for a given MW power. Figure 1a-d shows typical nanometerscale STM images of the HOPG surface following MW oxygen

Atomic Vacancy Engineering of Graphitic Surfaces

J. Phys. Chem. C, Vol. 113, No. 23, 2009 10251

plasma exposure at different powers and treatment times: 30 W, 4.5 s (a); 30 W, 6 s (b); 80 W, 4 s (c); and 150 W, 4 s (d). In all cases, the HOPG basal surface is decorated with a number of defects of uniform size, visualized as protrusions (bright spots), which are not present on pristine HOPG. The defect (spot) density is about 3 × 103 µm-2 (a), 2.5 × 105 µm-2 (b), 6 × 104 µm-2 (c), and 3 × 105 µm-2 (d). Figure 1e shows an atomic-scale STM image of the 80 W/4 s sample, where the individual defects, showing a locally distorted atomic structure, are more clearly noticed. Significantly, they all have the same appearance, with very similar heights (∼0.2-0.3 nm) and an extremely narrow diameter distribution having an average value of 0.8 ( 0.1 nm (Figure 1f), indicating that only one type of defect is present on the graphite surface. This result is in marked contrast to that obtained in the case of defects generated by electron or ion bombardment of graphite, where different types of defect structures and much wider defect diameter distributions are typically reported by STM.17,27,28 The defect generated by the present MW oxygen plasma must be ascribed to the single vacancy for the following reasons. First, the energy of the ions produced in the MW plasma is not enough to create defects on the HOPG surface by physical bombardment.25 Consequently, interstitial defects, which only appear as a result of energetic particle bombardment, cannot be the origin of the bright spots in Figure 1. Furthermore, an indication of the presence of interstitial atoms is the observation by STM of domelike defects with unperturbed lateral atomic arrangement,28 but this type of defect structure was never observed in the present case. Second, upon extended exposure of HOPG to the MW oxygen plasma, the small protrusions reported above develop first into larger protrusions (a few nanometers in diameter) and then into nanometer-sized, monolayer-deep pits.25,26 This implies that the attack progresses via abstraction of surface carbon atoms, so at a sufficiently early stage of the etching process only one carbon atom must have been removed from the site of the eventual pits, giving rise to single vacancies. The protrusions reported in Figure 1 are the first and smallest defects to appear on the HOPG surface following exposure to the MW oxygen plasma, so they must be attributed to the expected single vacancies. Third, a single vacancy on the graphite surface induces an enhancement of the local density of electronic states close to the Fermi level for the atoms surrounding the vacancy, so the vacancy is visualized as a protrusion (bright spot) in the STM images.27,28 In addition, theoretical calculations predict the lateral size for the single vacancy-induced protrusion in graphite and carbon nanotubes to be slightly below 1 nm,29,30 in agreement with the features observed in the present work (Figure 1). As shown below, the ability to generate a defect population of monodisperse type and tailored density on the HOPG surface, made up exclusively of single vacancies, provides a unique opportunity to directly probe the migration behavior of this type of defect. The migration of point defects, and in particular of the single vacancy, is a thermally activated process characterized by a migration energy Em. The number of atomistic jumps per second, ν, experienced by a vacancy during its diffusion follows an Arrhenius law, which is expressed as

ν ) ν0 exp(-Em /kT)

(1)

where ν0 ∼ 1013 s-1 is the pre-exponential factor, k is the Boltzmann constant, and T is the absolute temperature.18,19 In general, the diffusion of a vacancy on the graphite surface will proceed until (i) it is annihilated with a point defect of opposite

sign (i.e., a carbon interstitial) or with an extended crystallographic defect, such as a step edge or a grain boundary, or (ii) it encounters another vacancy, forming an aggregate (vacancy cluster) that reduces the free energy of the system and is therefore more stable than the two isolated single vacancies.18 Thus, the formation of vacancy clusters (e.g., through heat treatment) on an HOPG surface originally decorated just with isolated single vacancies constitutes direct proof of the migration of the latter. For aggregation to occur in a statistically significant number of cases, the density of single vacancies on the starting graphite surface should be relatively high. Otherwise, the probability for two vacancies to meet and form a cluster would be too low, so most of them would disappear by annihilation with the step edges or grain boundaries present on the graphite surface, and no or very few vacancy clusters would develop. Vacancy densities of about 6 × 104 µm-2 (Figure 1c and e) proved to be suitable for the mentioned purpose. A density of 6 × 104 µm-2 implies that approximately 0.15% of the carbon atoms from the surface graphene are missing and also that the typical separation between neighboring single vacancies is just a few nanometers (Figure 1c and e). When an HOPG surface densely decorated with single vacancies is subjected to heat treatment at a given temperature, T, for a given time, t, the single vacancies will migrate over the graphene sheet following random paths, and the distance migrated along such paths will be determined by the atomistic jump frequency of eq 1. For a fixed t, the distance migrated by the vacancies will increase with increasing heat treatment temperature. If the temperature is too low, the migration distance will be too short (i.e., clearly below the typical separation between vacancies) to allow neighboring vacancies to aggregate, so isolated single vacancies should be virtually the only feature imaged by STM on the graphite surface after the heat treatment. On the other hand, if the heat treatment temperature is sufficiently high, the migration distance will be long enough (i.e., clearly above the typical separation distance between vacancies) to allow vacancy aggregation, so vacancy clusters should be observed in the STM images. Thus, by performing a series of heat treatments to single vacancydecorated HOPG samples at different temperatures, we should be able to determine (via STM imaging) a range of temperatures T e Tl for which no vacancy aggregation occurs, and a range of temperatures T g Tu for which vacancy aggregation does take place. From the knowledge of the temperature Tl (Tu), using eq 1 and taking into account the aforementioned considerations, we can estimate a lower (upper) limit for the migration energy, l u l u l (Em ), so that Em g Em (Em e Em ), and finally Em e Em e Em u Em. For graphite samples with a vacancy density of 6 × 104 µm-2 (i.e., treated in MW plasma at a power of 80 W for 4 s), heat treatments at different temperatures and t ) 60 min were carried out under a flowing Ar atmosphere to ascertain the temperatures Tu and Tl. It was concluded that Tu ∼ 75 °C and Tl ∼ 50 °C. Figure 2a shows a typical nanometer-scale STM image of an 80 W/4 s sample subjected to heat treatment at 75 °C for 60 min. Before the heat treatment, the graphite surface was only decorated with isolated single vacancies (i.e., similar to Figure 1c and e). After the heat treatment, the development of bright features (∼0.3-0.4 nm high) of considerably larger lateral size than that characteristic of single vacancies became apparent (Figure 2a). Figure 2b shows an atomic-scale image of one of these large features, which appear to be constituted by several smaller parts, it being consistent with an aggregation of several different single vacancies. Figure 2c presents a

10252

J. Phys. Chem. C, Vol. 113, No. 23, 2009

Paredes et al. TABLE 1: Average Number of Atomistic Jumps Experienced by the Single Vacancy for Different Possible Migration Energies for 60 min at 50 and 75 °C (According to Equation 1) number of atomistic jumps in 60 min

Figure 2. (a, b) Representative nanometer- (a) and atomic-scale (b) STM images of a graphite surface originally decorated with isolated single vacancies (density ∼6 × 104 µm-2; see Figure 1c and e) and subjected to heat treatment at 75 °C for 60 min. (c) Histograms comparing the lateral size distribution of the bright features seen by STM on graphite surfaces originally decorated with isolated single vacancies (density ∼6 × 104 µm-2) before any heat treatment (blue) and after heat treatment at 75 °C (red) and 175 °C (yellow) for 60 min. The histogram ordinate represents normalized frequency. (d) Typical CP-AFM current image of a graphite surface with aggregated vacancies.

comparison of the diameter distributions of the bright features for HOPG surfaces originally decorated with isolated single vacancies (density about 6 × 104 µm-2) before any heat treatment (blue) and after heat treatment at temperatures of 75 °C (red) and 175 °C (yellow) for 60 min. After heat treatment, the proportion of features ascribed to isolated single vacancies (i.e., features with sizes mostly below 1 nm) decreases significantly, whereas features between 1 and 4 nm in size, attributed to the aggregation of single vacancies, become dominant. Such effect is more noticeable for the sample heat treated at 175 °C. The average feature size was determined to be 2.0 ( 1.0 and 2.8 ( 1.0 nm for the samples heat treated at 75 and 175 °C, respectively. This is to be compared with the value of 0.8 ( 0.1 nm for the same samples before heat treatment. Below 75 °C (e.g., at 50 °C), no significant vacancy aggregation was observed in the STM images, and the feature size distributions were virtually identical to that of the non-heat-treated samples with isolated single vacancies. Figure 2d shows a current image of a typical HOPG sample with aggregated vacancies, as obtained by conducting probe atomic force microscopy (CP-AFM). As opposed to the case of STM, in CP-AFM it is possible to discriminate between topographical and electrical information, although the resolution is not as high as that of STM.31 For this reason, it was possible to detect spatial variations in the CP-AFM current signal only for samples with aggregated vacancies, which in turn allowed one to unveil their electrical signature, but not for samples with isolated single vacancies. The CP-AFM topographic images of HOPG samples with aggregated vacancies (not shown) were essentially flat and featureless. By contrast, their simultaneously recorded current images clearly exhibited localized areas of increased current flow (bright spots in Figure 2d), not present in pristine, untreated HOPG. Therefore, they must be related to the bright features seen in the STM images (Figure 2a). These observations lend further support to the idea that such

migration energy (eV)

50 °C

75 °C

0.8 0.9 1.0 1.1

12052 332 9 0.25

94816 3384 121 4

features are brought about by the presence of atomic vacancies on the HOPG surface. It is well-known that atomic vacancies on graphite do not result in raised topographical features (they appear flat in AFM topography images), but only in an increased local density of electronic states near the Fermi level for the atoms surrounding the vacancy; i.e., they correspond to areas of enhanced local conductivity.27,28,32 Consequently, if the bright features seen in the STM images were the signature of atomic vacancies, they should appear flat in the CP-AFM topographical images and with bright contrast (increased current flow) in the corresponding current images, as it was indeed the case. From the experimentally determined value of Tu (∼75 °C), and making use of eq 1, a migration time of 60 min, and an atomistic jump distance of 0.142 nm (the C-C distance in graphite),19,20 we deduce that Em j 1.0 eV. Similarly, from the value of Tl (∼50 °C), we deduce that Em J 0.9 eV. From both inequalities, we can estimate the migration energy for the single vacancy as Em ≈ 0.9-1.0 eV. To make such estimation, we calculated the number of atomistic jumps that are typically required for the aggregation of two single vacancies randomly diffusing over the graphite surface. For a vacancy density of 6 × 104 µm-2, and assuming a two-dimensional random walk process, it was determined that between a few hundred and a few thousand atomistic jumps are typically required for two vacancies to aggregate. When we compare such figures with the average number of atomistic jumps for different migration energies at the two experimentally relevant temperatures (50 and 75 °C; see Table 1), we conclude that the migration energy must be ∼0.9-1.0 eV. A migration energy of 1.1 eV is not possible, because in a such case no aggregation should be observed at 75 °C, which contradicts the experimental observation. Likewise, a migration energy of 0.8 eV is also not possible, as aggregation should clearly take place at 50 °C, contradicting again the experiments. To all intents and purposes, the estimation of Em as ∼0.9-1.0 eV is reasonably accurate. This value is much smaller than the traditionally accepted experimental one (3.1 eV),1,18,19,21 which was obtained through TEM observations of the growth kinetics of microscopic vacancy clusters in irradiated graphite, but significantly lies within the range of values predicted by theory.1,19-21 At present, the most widely accepted theoretical value for Em is 1.7 eV.1,15,19 However, more recent calculations have yielded even lower values: 1.3-1.4 eV21 and 0.94-1.01 eV.20 Apart from broadly reconciling experimental observation with theory, the results reported in this work indicate that the single vacancy in graphite/graphene possesses a relatively high mobility. Furthermore, they strongly suggest that the large activation energy of 3.1 eV, usually ascribed to the single vacancy migration, must be attributed to other types of defects, as already pointed out in previous reports based on theoretical work.1,19 As the HOPG samples after the plasma treatments and before the heat treatments were exposed to the ambient air,

Atomic Vacancy Engineering of Graphitic Surfaces it is in principle possible that foreign atoms or chemical groups covalently attached to the dangling bonds of the carbon atoms in the single vacancies could be affecting their migration. However, several observations suggest that this question is not critical in the present case, and the conclusions reported above are not significantly affected by it. First, we note that if an atom/chemical group (e.g., H or OH) were covalently bound to one of the dangling carbon atoms of the single vacancy, then the migration of the vacancy would be seriously hampered. This is because migration would now involve not only overcoming the energy barrier to move one dangling carbon atom to the vacant site but also dissociating the, e.g., C-H or C-OH bond, which requires an energy of a few electronvolts. With such increased energy barrier, migration of the single vacancy in a graphite sample subjected to a temperature of 75 °C for 60 min would have not been possible. As a matter of fact, it is well-known that desorption of even the most labile oxygen functionalities chemisorbed onto carbon surfaces requires temperatures in excess of 200 °C.33 Likewise, desorption of H atoms chemisorbed onto a pristine graphene surface necessitates temperatures above 400 °C,34 so desorption of H atoms covalently attached to the dangling carbon atoms of single vacancies will also require at least similar temperatures. We can therefore conclude that, because migration on the single vacancy-decorated HOPG surfaces was indeed observed following heat treatment at 75 °C, the vacancies (or a large majority of them) must have been free of attached chemical groups, at least during the period of time that elapsed between the end of the plasma treatment and the end of the heat treatment, which was typically a few hours. The question that immediately arises is why the vacancies have remained free of attached chemical species, at least for a certain period of time, if they were kept under ambient conditions. We believe that this is due to the fact that covalent attachment of chemical groups at the single vacancy sites through adsorption of atmospheric molecules generally requires overcoming some energy barrier, as suggested by recent theoretical calculations.35 Thus, the attachment of chemical groups at the single vacancy sites would not be an instantaneous process, but a thermally activated one, and therefore would progressively take place over a given period of time, not immediately after exposure of the plasma-treated graphite sample to the ambient air. Indeed, our experimental observations were consistent with this: when HOPG samples decorated with single vacancies were stored under ambient air at room temperature, we noticed (through STM imaging) that aggregation occurred in several days, which would be consistent with the migration barrier estimated in this work, but after that time, and over observation periods of many weeks, further aggregation of the vacancies was apparently arrested. This strongly suggests that most of the single vacancies remain free of chemical groups for a few to several days, so their migration and aggregation are possible during such a period, but after that time their further migration becomes arrested as chemical groups covalently attach to a large fraction of the vacancies. As most of the migration experiments reported here were carried out within a period of hours after preparation of the vacancy-decorated samples, we conclude that most of the vacancies were free of attached chemical groups, and the estimation of the migration barrier was not affected by this question. Another question that is worth discussing concerns the exact nature of the features observed when the single vacancy-

J. Phys. Chem. C, Vol. 113, No. 23, 2009 10253 decorated graphite surfaces are subjected to heat treatment at 75 °C (Figure 2). We believe that such features correspond to vacancy aggregates, i.e., to a number of single vacancies sitting very close to each other, rather than to coalesced vacancies (i.e., actual double-, triple-, or multivacancies). Theoretical calculations on graphene layers by Lee et al. indicate that when two single vacancies are sitting side by side, but still remaining as individual entities (i.e., not forming a double vacancy), the total energy of the system is reduced by a few electronvolts compared to the system with the two vacancies further apart, and also that the coalescence of the two side-by-side single vacancies into a double vacancy requires overcoming an energy barrier of 1.5-1.6 eV.20 This indicates that when two randomly diffusing single vacancies come to close proximity, they form a more or less stable aggregate (of two side-by-side single vacancies) at 75 °C, but they do not coalesce into a double vacancy, as the energy barriers required to move the two single vacancies away from each other or to coalesce them into a double vacancy would both be too high to be overcome at 75 °C. We also note that, according to theoretical calculations performed on carbon nanotubes,36 vacancies can interact with each other through strain fields, and this could in principle affect their migration. However, such interaction only becomes significant so as to affect vacancy diffusion when the two vacancies are just a couple of lattice parameters apart (i.e., below 1 nm), and we believe that at such small distance the two vacancies will have already formed the type of aggregate that we visualize here by STM and will not move away from each other. Therefore, this question will not affect the results and conclusions discussed here. Vacancy diffusion on the graphite surface could be induced not only on the whole sample by means of heat treatment but also locally through interaction with the STM tip. Most of the STM images reported in this work (including those of Figures 1 and 2) were obtained with a bias voltage of 100 mV and a tunneling current of 0.5-2.0 nA. Under such tunneling conditions, the small bright features attributed to isolated single vacancies appeared consistently at the same relative positions over many consecutive STM scans, implying that they were stable enough to withstand interaction with the STM tip. However, it was also observed that if the bias voltage and tunneling current were respectively reduced below and increased above a certain value, then vacancy diffusion could be induced on a local level. Figure 3a shows a general STM image (200 nm in lateral size) recorded at 100 mV and 0.5 nA, which corresponds to an HOPG sample with a density of single vacancies of about 3 × 105 µm-2, prepared by MW plasma treatment at 150 W for 4 s. As Figure 3a is a relatively low resolution image, the single vacancies cannot be individually resolved (see Figure 1d for a higher resolution image of the same sample). After this image was recorded, the STM scan size was reduced to 50 nm and the bias voltage and tunneling current were changed to 5 mV and 20 nA, respectively, so that the central area of Figure 3a was rescanned under such conditions. Then, the same 200 nm region was scanned again with the original tunneling parameters (100 mV and 0.5 nA), the result of which is depicted in Figure 3b. It can be noticed that the central 50 nm × 50 nm region has been modified and now appears decorated with bright features a few to several nanometers large (Figure 3c). Such observation leads us to conclude that vacancy diffusion and aggregation have been locally induced as a result of strong interactions between the STM tip and the graphite surface. It is well-known that

10254

J. Phys. Chem. C, Vol. 113, No. 23, 2009

Paredes et al.

Figure 3. (a) General STM image of a single vacancy-decorated graphite surface (density ∼3 × 105 µm-2) prepared by MW oxygen plasma treatment at 150 W for 4 s. Tunneling parameters: 100 mV (bias voltage) and 0.5 nA (tunneling current). (b) STM image of the same region recorded with the same tunnelling parameters as in (a) after scanning the central 50 nm × 50 nm region at 5 mV (bias voltage) and 20 nA (tunneling current). (c) Magnification of the 50 nm × 50 nm region as outlined by the red square in (b).

significant tip-sample force interactions come into play under normal STM operating conditions, and such force increases as the tunneling resistance is decreased.37 It was determined that, to induce vacancy aggregation, the bias voltage and tunneling current had to be set below 10 mV and above 10 nA, respectively. Although the exact mechanism that drives the tip-induced migration of the vacancies is not known at present, we propose that the strong forces exerted by the STM tip on the HOPG surface under rather severe tunneling parameters bring about a relaxation of the graphite lattice37 with a subsequent weakening of the carbon-carbon bonds, which in turn results in a local reduction in the effective migration energy of the single vacancy. Provided that the force exerted by the STM tip is strong enough, the effective migration energy of the vacancy will be reduced below the limit that allows its diffusion and aggregation at room temperature within the time scale of the STM scan (∼1 min). In such a case, using eq 1 we can estimate that the effective migration energy of the vacancy will be locally reduced to a value of ∼0.7 eV or below. A similar mechanism has been invoked to explain the tip-induced manipulation of single atoms on semiconductor surfaces using dynamic force microscopy.38 An alternative explanation would be that, because at a reduced tunneling resistance (i.e., 5 mV and 20 nA in the present case) the tip-sample separation is reduced significantly, the locally deposited energy by the tunnelling current will be considerably higher, which would favor the local annealing effects. In any case, the possibility of inducing vacancy migration on a local scale with the STM tip could be employed to control the spatial distribution of this type of defect (e.g., to pattern vacancy clusters at predefined locations) on graphite, graphene, and related carbon nanostructures, which has not been previously realized. This point is currently the focus of ongoing research in our laboratory. 4. Conclusions In conclusion, we have presented a plasma-based chemical approach for the selective introduction of single vacancies with tunable density on graphitic surfaces. As a first direct application of this possibility, the migration behavior of the single vacancy has been investigated. An energy barrier of ∼0.9-1.0 eV to the migration of this defect has been determined by means of scanning tunneling microscopy (STM) observations, which is much lower than the generally accepted experimental value (3.1 eV) but broadly agrees with theoretical calculations. Moreover, such energy barrier can be locally reduced to about 0.7 eV through interaction with the STM tip. Concerning prospective applications of the proposed approach, since plasma etching is scalable and compatible with microfabrication technologies, it could be highly useful, e.g., in future graphene-based electronic

devices. For example, required densities of single vacancies could be patterned onto preselected areas of graphene sheets or nanoribbons to impart a given functionality to the electronic device.2 Acknowledgment. P.S.-F. acknowledges an I3P predoctoral contract from CSIC. Partial funding of this work by the Spanish MEC through projects CTQ2004-07698-C02-02 and CTQ200509105-C04-02 is gratefully acknowledged. References and Notes (1) Telling, R. H.; Ewels, C. P.; El-Barbary, A. A.; Heggie, M. I. Nat. Mater. 2003, 2, 337–337. (2) Krasheninnikov, A. V.; Banhart, F. Nat. Mater. 2007, 6, 723–733. (3) Ajima, K.; Yudasaka, M.; Suenaga, K.; Kasuya, D.; Azami, T.; Iijima, S. AdV. Mater. 2004, 16, 397–401. (4) Robinson, J. A.; Snow, E. S.; Baˇdescu, S. C.; Reinecke, T. L.; Perkins, F. K. Nano Lett. 2006, 6, 1747–1751. (5) Suenaga, K.; Wakabayashi, H.; Koshino, M.; Sato, Y.; Urita, K.; Iijima, S. Nat. Nanotechnol. 2007, 2, 358–360. (6) Osva´th, Z.; Verte´is, G.; Tapaszto´, L.; We´ber, F.; Horva´th, Z. E.; Gyulai, J.; Biro´, L. P. Phys. ReV. B 2005, 72, 045429. (7) Tapaszto´, L.; Dobrik, G.; Nemes-Incze, P.; Verte´is, G.; Lambin, Ph.; Biro´, L. P. Phys. ReV. B 2008, 78, 233407. (8) Go´mez-Navarro, C.; de Pablo, P. J.; Go´mez-Herrero, J.; Biel, B.; Garcia-Vidal, F. J.; Rubio, A.; Flores, F. Nat. Mater. 2005, 4, 534–539. (9) Amorim, R. G.; Fazzio, A.; Antonelli, A.; Novaes, F. D.; da Silva, A. J. R. Nano Lett. 2007, 7, 2459–2462. (10) Zhang, S.; Mielke, S. L.; Khare, R.; Troya, D.; Ruoff, R. S.; Schatz, G. C.; Belytschko, T. Phys. ReV. B 2005, 71, 115403. (11) Terrones, M.; Terrones, H.; Banhart, F.; Charlier, J. C.; Ajayan, P. M. Science 2000, 288, 1226–1229. (12) Kostov, M. K.; Santiso, E. E.; George, A. M.; Gubbins, K. E.; Buongiorno Nardelli, M. Phys. ReV. Lett. 2005, 95, 136105. (13) Han, K.-H.; Spemann, D.; Esquinazi, P.; Ho¨hne, R.; Riede, V.; Butz, T. AdV. Mater. 2003, 15, 1719–1722. (14) Lehtinen, P. O.; Foster, A. S.; Ma, Y.; Krasheninnikov, A. V.; Nieminen, R. M. Phys. ReV. Lett. 2004, 93, 187202. (15) Hashimoto, A.; Suenaga, K.; Gloter, A.; Urita, K.; Iijima, S. Nature 2004, 430, 870–873. (16) Yazyev, O. V.; Tavernelli, I.; Rothlisberger, U.; Helm, L. Phys. ReV. B 2007, 75, 115418. (17) Pregler, S. K.; Hayakawa, T.; Yasumatsu, H.; Kondow, T.; Sinnott, S. B. Nucl. Instrum. Methods B 2007, 262, 240–248. (18) Banhart, F. Rep. Prog. Phys. 1999, 62, 1181–1221. (19) El-Barbary, A. A.; Telling, R. H.; Ewels, C. P.; Heggie, M. I.; Briddon, P. R. Phys. ReV. B 2003, 68, 144107. (20) Lee, G.-D.; Wang, C. Z.; Yoon, E.; Hwang, N.-M.; Kim, D.-Y.; Ho, K. M. Phys. ReV. Lett. 2005, 95, 205501. (21) Krasheninnikov, A. V.; Lehtinen, P. O.; Foster, A. S.; Nieminen, R. M. Chem. Phys. Lett. 2006, 418, 132–136. (22) Chang, H.; Bard, A. J. J. Am. Chem. Soc. 1991, 113, 5588–5596. (23) Nicholson, K. T.; Minton, T. K.; Sibener, S. J. J. Phys. Chem. B 2005, 109, 8476–8480. (24) Tracz, A.; Wegner, G.; Rabe, J. P. Langmuir 2003, 19, 6807–6812. (25) Paredes, J. I.; Martı´nez-Alonso, A.; Tasco´n, J. M. D. Langmuir 2002, 18, 4314–4323. (26) Paredes, J. I.; Martı´nez-Alonso, A.; Tasco´n, J. M. D. Langmuir 2007, 23, 8932–8943. (27) Hahn, J. R.; Kang, H. Phys. ReV. B 1999, 60, 6007–6017.

Atomic Vacancy Engineering of Graphitic Surfaces (28) Kibsgaard, J.; Lauritsen, J. V.; Lægsgaard, E.; Clausen, B. S.; Topsøe, H.; Besenbacher, F. J. Am. Chem. Soc. 2006, 128, 13950–13958. (29) Krasheninnikov, A. V.; Elesin, V. F. Surf. Sci. 2000, 454-456, 519–524. (30) Krasheninnikov, A. V.; Nordlund, K.; Sirvio¨, M.; Salonen, E.; Keinonen, J. Phys. ReV. B 2001, 63, 245405. (31) Palermo, V.; Liscio, A.; Palma, M.; Surin, M.; Lazzaroni, R.; Samorı´, P. Chem. Commun. 2007, 3326–3337. (32) Hahn, J. R.; Kang, H.; Song, S.; Jeon, I. C. Phys. ReV. B 1996, 53, R1725–R1728. (33) Figueiredo, J. L.; Pereira, M. F. R.; Freitas, M. M. A.; Orfao, J. J. M. Carbon 1999, 37, 1379–1389.

J. Phys. Chem. C, Vol. 113, No. 23, 2009 10255 (34) Elias, D. C.; Nair, R. R.; Mohiuddin, T. M. G.; Morozov, S. V.; Blake, P.; Halsall, M. P.; Ferrari, A. C.; Boukhvalov, D. W.; Katsnelson, M. I.; Geim, A. K.; Novoselov, K. S. Science 2009, 323, 610–613. (35) Allouche, A.; Ferro, Y. Carbon 2006, 44, 3320–3327. (36) Kotakoski, J.; Krasheninnikov, A. V.; Nordlund, K. Phys. ReV. B 2006, 74, 245420. (37) Magonov, S. N.; Whangbo, M.-H. Surface Analysis with STM and AFM; VCH: Weinheim, 1996; Chapter 8. (38) Sugimoto, Y.; Jelinek, P.; Pou, P.; Abe, M.; Morita, S.; Perez, R.; Custance, O. Phys. ReV. Lett. 2007, 98, 106104.

JP901578C