In Situ TEM Study on the Electrical and Field-Emission Properties of

Oct 18, 2012 - Key Laboratory of Interface Science and Engineering in Advanced Materials, Taiyuan University of Technology, Taiyuan 30024, China. J. P...
0 downloads 0 Views 1MB Size
Article pubs.acs.org/JPCC

In Situ TEM Study on the Electrical and Field-Emission Properties of Individual Fe3C‑Filled Carbon Nanotubes Qingmei Su,† Jie Li,† Gaohui Du,*,†,‡ and Bingshe Xu‡ †

Institute of Physical Chemistry, Zhejiang Normal University, Jinhua 321004, China Key Laboratory of Interface Science and Engineering in Advanced Materials, Taiyuan University of Technology, Taiyuan 30024, China



ABSTRACT: The electrical and electron field-emission characteristics of individual Fe3C-filled CNTs were investigated using a scanning tunneling microscope inside a transmission electron microscope. Larger bias sweeping across a CNT can bring about the melting, resolidifying, and decomposition of the encapsulated Fe3C, which causes structural damage to the CNT. However, the resistance of the CNT reduces significantly after large bias sweep because the graphitization of CNT walls improves and the contact resistance reduces. To get more insights into the mechanism, the microstructure evolution of Fe3C and the reaction between Fe3C and CNT during near-equilibrium reaction processes were studied. It was found that thin graphenes form when the Fe3C fillers decompose by joule heating, which may open up a new route for the graphene fabrication. The field-emission behavior of a single Fe3C-filled CNT was analyzed based on Fowler−Nordheim theory, which suggested only 8.4% of the hemispherical cap is responsible for the electron emission. length, and number of shells of isolated CNTs.16 In-situ TEM measurements proved that the transport characteristics of double-walled CNTs can be directly correlated with their chiral indices.17 Peng et al. found that multiple CNTs can be soldered together with moderate junction resistance in TEM, and the resistance may be further reduced by current-induced graphitization;18 they also reported that long time fieldemission usually results in drastic structure damage that may lead to sudden emission failure,19 and the electron fieldemission properties of CNTs were highly sensitive to the emission tip structures.20 Since iron is a widely used catalyst for the CNT preparation, iron carbides are prone to form during the CNT growth. So Fe3C nanoparticles were frequently observed in CNT products.21,22 It is not easy to eliminate these Fe3C nanoparticles to get pure CNTs because they are often encapsulated within the cavities of CNTs. So it brings about a question for researchers: how the Fe3C nanostructures influence the electrical and field-emission properties of CNTs in future applications. This paper might shed light on this question by performing a comprehensive in situ study on the structures, electrical and field-emission properties of individual Fe3C-filled CNTs inside a TEM with a STM-TEM system.

1. INTRODUCTION The characterizations of mechanical, electrical, and chemical properties of carbon nanotubes (CNTs) become greatly important because of their special structures, remarkable properties and potential applications in nanodevices.1−3 One of the most exciting and promising applications of CNTs is in the field of nanoelectronics.4 Recently, the fascinating electronic properties of CNTs can be varied from metal to semiconducting by their atomic structures,5 and introducing foreign materials into the hollow cavities of CNTs in a confined configuration would cause a significant change in the electrical properties of the CNTs.6 The CNTs filled with various nanostructures have been expected to have many applications, such as microwave absorption,7 catalysts, spectroscopic enhancers, and flat field-emission display.8,9 In particular, the CNT composite is considered as one of the most promising electron field emitters due to the excellent geometric, electrical and mechanical properties.10 The advent of nanotechnology has motivated decisive progresses in characterization methods with high spatial resolution. The recent surge of in situ TEM, especially the combination of scanning tunneling microscopy (STM) and transmission electron microscopy (TEM), has evolved into a powerful tool for research on the property measurement of nanomaterials.11−15 Electrical property measurements of nanostructures with time-resolved high spatial and chemical resolution using a TEM holder with integrated STM units have attracted great attention recently.12−15 Kociak et al. presented the measurements of the transport properties accompanying with the determination of the structure, including chiral indices, © 2012 American Chemical Society

Received: September 13, 2012 Revised: October 16, 2012 Published: October 18, 2012 23175

dx.doi.org/10.1021/jp309121f | J. Phys. Chem. C 2012, 116, 23175−23179

The Journal of Physical Chemistry C

Article

2. EXPERIMENTAL SECTION 2.1. Preparation of Fe3C-Filled CNTs. Fe3C-filled CNTs were fabricated in a conventional thermal CVD system consisting of a horizontal furnace fitted with a quartz tube. A Si wafer was placed in the quartz tube as a substrate to deposit the CNTs. The CVD furnace was heated up to the reaction temperature of 1000 °C with the flow of Ar (2000 sccm) and H2 (200 sccm) in the reactor. The ferrocene−ethanol solution was injected in the quartz tube reactor through a syringe to initiate the CNT growth. After reaction for 30 min, the CVD furnace was cooled down naturally to room temperature with the purge of Ar (200 sccm). Upon completion of the experiment, a black CNT film was formed on the Si substrate. The ferrocene-ethanol solution was prepared as follows: 1 g of ferrocene and 3 g of dichlorobenzene were dissolved in 50 mL of ethanol; then, the mixture was ultrasonicated for 20 min to obtain a clear orange solution. 2.2. Characterization. The morphology, composition and microstructures of the products were analyzed using powder Xray diffraction (XRD, Cu Kα radiation, Philips PW3040/60) and high-resolution transmission electron microscope (TEM, JEOL 2100F) . Electron energy-loss spectroscopy (EELS) experiment was performed using the Gatan Enfina EELS system attached to the JEOL 2100F. The in situ TEM experiments were carried out using a “Nanofactory Instruments AB” STM-TEM holder inserted into a JEM-2010F TEM. A sharp tungsten STM tip was prepared by electrochemical etching of tungsten wires with a diameter of 0.25 mm in 5 M NaOH solution; a cut gold wire was used as counter-electrode. CNTs sample was attached on the gold electrode by rubbing the Au-tip on the CNT film. The tungsten and gold tips were then mounted in the STM-TEM holder, and the STM probe can be positioned with subnanometer resolution with the STM unit actuated by a pizeotube, making it possible to select a specific CNT to perform an electrical or field emission measurements inside the TEM.

Figure 1. (a) XRD pattern, (b, c) TEM images, and (d) EELS spectrum of the CNTs. (e) HRTEM image of a Fe3C nanowire-filled CNT and (f) its corresponding FFT pattern. (g) HRTEM image of another filled CNT and (h) its corresponding SAED pattern.

fringes of the Fe3C crystal, respectively. The growth direction of the Fe3C within the CNT is determined to be along [100] direction. XRD and TEM analysis demonstrate the product is pure Fe3C-filled CNTs. A single Fe3C-filled CNT was selected in TEM by moving the W tip to contact with it to perform a two terminal I−V measurements. The selected CNT is about several micrometers filled with several segments of Fe3C; only a part of the CNT was in situ monitored in TEM. Figure 2 shows a serial of TEM images of the Fe3C-filled CNT after the fast sweep of different bias, revealing the changes of the microstructure and electrical properties of the Fe3C-filled CNT. TEM images were recorded after each sweep with voltage from −x to x volt (x ranges from 0.1 to 2 V with an increase of 0.1 V each time) applied to the two electrodes. When the sweep voltage is less than 0.8 V, this is no structural change observed for the CNT (Figure 2a), and the resistance remained at ∼32.1 kΩ. Figure 2b gives a TEM image of the CNT after a sweep of 0.9 V with the maximum instant current of 43.58 μA; the filled Fe3C began melting during the bias sweep. A Fe3C segment below the viewing image melted quickly and flowed to form a long segment by soldering with the segment in viewing area. The CNT walls did not show evident change. The corresponding I−V curve is shown in Figure 2g by a black line, from which a resistance of ∼20.65 kΩ was obtained, revealing a decrease of 36%. We think this decrease can be ascribed to the improvement of graphitization due to high temperature by Joule heating. Figure 2c gives a TEM image of the CNT after a sweep of 1.4 V; with

3. RESULTS AND DISCUSSION XRD analysis and TEM images of the as-grown CNTs are shown in Figure 1. Shown in Figure 1a is the XRD pattern of the sample. The main diffraction peaks can be assigned to orthorhombic iron carbide (a = 5.089, b = 6.744, c = 4.524, JCPDS card No. 85-1317) and graphite (JCPDS card No. 752078). The results suggested that the sample was possibly Fe3C-filled CNTs, and it was confirmed by the following TEM analysis. Figure 1b is a TEM image of the obtained CNTs showing a uniform diameter round 50 nm. It is obvious that many of the CNTs are filled with other material. The fillers can be several segments or long nanowires (Figure 1c). The EELS spectrum recorded from a filled CNT is shown in Figure 1d, indicating it is made of C (C-K: 283.8 eV) and Fe (Fe-L2: 721 eV; Fe-L3: 708 eV). The microstructure analysis was further performed by means of high-resolution TEM (HRTEM). Figure 1e,f is a HRTEM image of a nanowire-filled CNT and its corresponding fast Fourier transform (FFT) pattern. The boundary between CNT and the filling material is clear; the fringe spacing of 0.20 and 0.22 nm are observed in the HRTEM image, corresponding to the (112) and (21̅1̅) planes of orthorhombic Fe3C. Figure 1g,h is a HRTEM image of another filled CNT and its corresponding selected-area electron diffraction (SAED) pattern, in which the measured spacing of 0.68 and 0.25 nm corresponds to the (010) and (200) lattice 23176

dx.doi.org/10.1021/jp309121f | J. Phys. Chem. C 2012, 116, 23175−23179

The Journal of Physical Chemistry C

Article

bias of 0.9 V, which should be a result of the improvement of graphitization degree due to high temperature by Joule heating and the instant reaction of Fe3C with CNT. The resistance continues to reduce gradually with a sweeping bias between 1.0 and 1.7 V, which can be attributed to the decrease of the contact resistance between the CNT and the probe and the further improvement of graphitization within CNT. After the sweeping of 1.7 V, the resistance tends toward a stable value. The I−V curves shown in Figure 2g suggest a lineal I−V feature at low bias sweep (the black line) and a nonlineal I−V behavior for higher bias sweep (the red and green lines), same with the previous report.23 To reveal the microstructure evolution of Fe3C and the reaction between Fe3C and CNT during fast voltage sweep, the bias applied to a CNT was slowly increased in the experiments. Therefore, the near- equilibrium process can be observed under gradual Joule heating. Figure 3 shows a time sequence of TEM

Figure 2. Serial of TEM images of a Fe3C-filled CNT after the different bias sweep: (a) 0 V, (b) 0.9 V, (c) 1.4 V, (d) 1.7 V, and (e) 2 V. (f) SAED pattern recorded from a Fe3C particle in part e. (g) Corresponding I−V curves for parts b, c, and e are shown by black, red, and green lines, respectively. (h) Resistance distributions with increasing sweeping bias.

the increasing of bias, an obvious change was observed at the end of CNT, namely the contacting part of the CNT with the W electrode. The forepart of the filled Fe3C nanorods could be decomposed and a few graphite structures were formed. The change should be resulted from the Joule heating due to higher contact resistance. The corresponding I−V curve is shown in Figure 2g by the red line, from which a resistance of ∼15.68 kΩ was obtained. When the sweep voltage increased to 1.7 V with the maximum instant current of 100.84 μA (Figure 2d), a huge change occurred to the CNT: (1) the contacting part of the CNT was broken and formed a better contact; (2) the filling nanorods melted and resolidified to form particles, and the CNT walls showed certain destruction. After the sweep of 2 V with an instant maximum current of 155.04 μA, major of the filled Fe3C decomposed and evaporated, and just a few nanoparticles were left (Figure 2e). The CNT walls were further destructed due to the reaction with Fe3C. A SAED pattern was recorded from a single nanoparticle, and suggested that the particle was still Fe3C (Figure 2f). The I−V curve for the CNT with 2 V sweep is shown in Figure 2g by green line, from which the calculated resistance was 12.90 kΩ. Figure 2h gives the resistance variation with different bias sweep, which decreases from 32.1 to 12.90 kΩ in the whole electrical measurement. There is no structural change when the bias is lower than 0.9 V (the maximum instant current less than 43.58 μA). The resistance reduces dramatically at the sweep

Figure 3. TEM images of a single Fe3C-filled CNT (a) and its microstructure evolution (b−i) with an applied bias of 2137 mV. The scale bars are 20 nm. (j) The resistance variation of this CNT at different stage in parts a−i. (k) The mass lost of the Fe3C particle at different stage in parts a−i.

video images of the reaction process. Figure 3a shows the initial stage of a Fe3C nanoparticles-filled CNT. The Fe3C nanoparticle has a diameter of approximate 37 nm and a length of 135 nm; therefore, we determined the mass of the original Fe3C is approximately 1.11 fg according to the density of cementite (7.649 g/cm3). Then a bias was applied to the two ends of the CNT by the W and Au probes, and the voltage was slowly increased from 0 mV. When the voltage reached 2137 mV with 67 μA current, we found the filling Fe3C moved about 78 nm suddenly, indicating the reaction began. It was obvious that some graphite structures formed along the moving trajectory of Fe3C nanoparticle (Figure 3b). When the nanoparticle was brought to the viewing area, we found it has shrunk notably (Figure 3c). The carbide nanoparticle decrease to 49.2 nm in length, correspondingly to the mass of ∼0.41 fg, suggesting that 23177

dx.doi.org/10.1021/jp309121f | J. Phys. Chem. C 2012, 116, 23175−23179

The Journal of Physical Chemistry C

Article

the carbide began decomposing. The continuous decomposition was taking place inside the CNT with the fixed bias voltage at 2137 mV as shown in Figure 3d−i. The first to be noticed was the decreasing size; eventually, the filling section completely disappeared from the inner cavity of the CNT. Accompanying with this process, the shells of the CNT were destructed locally due to the reaction with Fe3C particle. The significant feature is the formation of new graphite nanostructures following the tail of the Fe3C particle, which can be clearly seen in Figure 3e−g. Another noticeable feature is the electromigration phenomenon; in other words, the Fe3C particle was gradually moved along the direction to anode during the whole reaction process. In a previous report,24 the structure transform can be achieved by Joule heating and strong electron beam bombardment in TEM. Electron beam bombardment should not be the main mechanism in this experiment because of low intensity of the beam and short irradiation time. Therefore, we attribute transportation current-induced Joule heating to be the main factor responsible for the Fe3C decomposing and reacting processes. The Fe3C decomposed to form Fe and C atoms during Joule heating while the C atoms were quickly graphitized into fine graphite nanostructures. Fe atoms could be evaporated or electro-migrated because no Fe nanoparticle was observed after reaction. Figure 3j shows the resistance of this single CNT at different reaction stage. The resistance displays slight decrease and the reason should be attributed to the appearance of the fine graphite nanostructures with better graphitization degree. The total change of the resistance is not remarkable because this Fe3C particle is small; the attribution of the formed fine graphite structure to the conductance of the whole CNT is less compared with the Fe3C nanowire-filled CNT in Figure 2. Accordingly, the estimated mass lost corresponding to each image is shown in Figure 3k. The entire reaction process consumed about 5.5 min, and the average mass lost is about 0.21 fg/min. So the mass lost can be well controllable, allowing precise processing of the filler at the femtogram scale for timebased control. The structural evolution of Fe3C nanoparticle and its reaction with CNT shells have been demonstrated in Figure 3. Since the Fe3C nanoparticle was encapsulated within a CNT, the decomposing process itself was not clearly revealed. To investigate this process, a Fe3C nanoparticle attached to a CNT tip was monitored during gradual Joule heating. A series of TEM images are recorded and shown in Figure 4. In Figure 4a, the initial Fe3C spherical nanoparticle is approximately 54 nm in diameter. Its volume reduced gradually as shown in Figure 4b−h, and the particle disappeared eventually in Figure 4i. We can see the growth of graphene sheet around the particle accompanying with the volume shrinkage, as indicated by an arrowhead in Figure 4c. Furthermore, thin corrugated graphene sheets formed when the Fe3C particle decomposed completely (Figure 4i). This process suggests a new route to prepare graphene by thermally decomposing iron carbides; on the other hand, it explained why the conductance of Fe3C-filled CNT improves after larger bias sweeping while the CNT shells seem corrupted. The electron field-emission properties of a single Fe3C nanowire-filled CNT were measured in situ in TEM with the W tip as the anode (as shown in Figure 5a). In our experiments the W electrode was stationary while the CNT was moved to adjust the relative position between the CNT-tip and W anode. Figure 5b shows the field-emission current versus voltage (I−

Figure 4. (a−i) Collection of representative video frames of the decomposing process of a Fe3C nanoparticle attached to a CNT tip. The arrowheads in images reveal the formation of corrugated graphene sheets. Scale bars: 20 nm.

V) curve for the CNT. Usually the Fowler−Nordheim (FN) law is used to analyze the field-emission I−V measurements, which gives a relationship between the electron emission current and the local field at the emitter surface. The F−N equation can be written as25 I=A

2 ⎛ 10.4 ⎞ 1.5 × 10−6 ⎛⎜ V ⎞⎟ 2 ⎟ γ exp⎜ ⎝d⎠ φ ⎝ φ⎠

⎛ (6.44 × 109)φ15d ⎞ exp⎜ − ⎟ γV ⎠ ⎝

where A is the effective emission area, ϕ is the work function of the emitter, d is the distance between the electrodes, and γ is the field enhancement factor . The method to determine A and γ from the measurements is to make a F−N plot, i.e., to plot ln(I/V2) versus 1/V, and this plot should appear as a straight line with its slope depending on d, ϕ, and γ. Because d can be directly determined from TEM image, γ can be calculated by taking a reasonable value for ϕ (5 eV for CNT). Figure 5b shows a field-emission I−V curve, and the F−N plot was inserted in the figure. The F−N plot is fitted well with a straight line over a wide range of the electric field, indicating the electron field-emission from the Fe3C-filled CNT follows the F−N law. The onset voltage (1 nA) of this Fe3Cfilled CNT is found to be 61.1 V. From the slope of the F−N plot we can obtain the field enhancement factor γ = 28, which is close to the previous measurement results on individual CNTs.26 The effective emission area obtained from the intercept of the F−N plot is 256.8 nm2, which is much smaller than the surface area of the hemispherical CNT cap (∼3041 nm2), suggesting that only a small part (8.4%) of the cap contributes effectively to the emission current. No structural change was observed for the CNT tip during field-emission measurement, indicating it can be used as a stable field emitter.

4. CONCLUSIONS In summary, the electrical properties of individual Fe3C-filled CNTs have been investigated using in situ TEM. It was revealed that the Fe3C nanostructures could melt, resolidify, and decompose under larger bias sweep, which caused 23178

dx.doi.org/10.1021/jp309121f | J. Phys. Chem. C 2012, 116, 23175−23179

The Journal of Physical Chemistry C

Article

Figure 5. (a) TEM image showing the CNT tip and W electrode. (b) Field-emission I−V curve and the corresponding F−N plot (the inset). The distance between the CNT tip and the electrode was 450 nm. (8) Ho, Y.; Tang, J.; Zhang, H. B.; Qian, C.; Feng, Y. Y.; Liu, J. ACS Nano 2009, 3, 1057−1062. (9) Soldano, C.; Rossella, F.; Bellani, V.; Giudicatti, S.; Kar, S. ACS Nano 2010, 4, 6573−6578. (10) De Heer, W. A.; Chatelain, A.; Ugarte, D. A. Science 1995, 270, 1179−1180. (11) Golberg, D.; Costa, P. M. F. J.; Mitome, M.; Bando, Y. Nano Res. 2008, 1, 166−175. (12) Costa, P. M. F. J.; Golberg, D.; Shen, G. Z.; Mitome, M.; Bando, Y. J. Mater. Sci. 2008, 43, 1460−1470. (13) Costa, P. M. F. J.; Gautam, U. K.; Bando, Y.; Golberg, D. Nat. Commun 2011, 1429, 1−6. (14) Golberg, D.; Costa, P. M. F. J.; Wang, M. S.; Wei, X. L.; Tang, D. M.; Xu, Z.; Huang, Y.; Gautam, U. K.; Liu, B. D.; Zeng, H. B.; et al. Adv. Mater. 2012, 24, 177−194. (15) Zhao, J.; Huang, J. Q.; Wei, F.; Zhu, J. Nano Lett. 2010, 10, 4309−4315. (16) Kociak, M.; Suenaga, K.; Hirahara, K.; Saito, Y.; Nakahira, T.; Iijima, S. Phys. Rev. Lett. 2002, 89, 155501. (17) Liu, K. H.; Wang, W. L.; Xu, Z.; Bai, X. D.; Wang, E. G.; Yao, Y. G.; Zhang, J.; Liu, Z. F. J. Am. Chem. Soc. 2009, 131, 62−63. (18) Wang, M. S.; Wang, J. Y.; Chen, Q.; Peng, L. M. Adv. Funct. Mater. 2005, 15, 1825−1831. (19) Jin, C. H.; Wang, J. Y.; Wang, M. S.; Su, J.; Peng, L. M. Carbon 2005, 43, 1026−1031. (20) Wang, M. S.; Wang, J. Y.; Peng, L. M. Appl. Phys. Lett. 2006, 88, 243108. (21) Su, Q. M.; Zhong, G.; Li, J.; Du, G. H.; Xu, B. S. Appl. Phys. A: Mater. Sci. Process. 2012, 106, 59−65. (22) Morelos-Gómez, A.; López-Urías, F.; Muñoz-Sandoval, E.; Dennis, C. L.; Shull, R. D.; Terrones, H.; Terrones, M. J. Mater. Chem. 2010, 20, 5906−5914. (23) Huang, J. Y.; Chen, S.; Jo, S. H.; Wang, Z.; Han, D. X.; Chen, G.; Dresselhaus, M. S.; Ren, Z. F. Phys. Rew. Lett 2005, 94, 236802. (24) Wu, J. H.; Han, L. B.; Wang, N.; Song, Y. L.; Chen, H. H.; Chen, H. H.; Hu, J. Q. CrystEngComm 2011, 13, 4611−4616. (25) Nilsson, L.; Groening, O.; Groening, P.; Kuettel, O.; Schlapbach, L. J. Appl. Phys. 2011, 90, 768−780. (26) Wang, M. S.; Peng, L. M.; Wang, J. Y.; Jin, C. H.; Chen, Q. J. Phys. Chem. B 2006, 110, 9397−9402.

structural damage to the CNT. However, the resistance of the CNT reduced after the bias sweep because of the improvements of the CNT graphitization and the contact resistance. To get insight into the mechanism, the microstructure evolution of the filling Fe3C and the reaction between Fe3C and CNT during near-equilibrium reaction processes were studied. It was demonstrated that thin corrugated graphene sheets form when Fe3C decompose under Joule heating; the formation of fine graphene nanostructures within a CNT accounts for the improvement of the conductance. This finding may also open up a new route for the graphene fabrication by thermally decomposing iron carbides. Moreover, the field-emission behavior of a single Fe3C nanowire-filled CNT was studied inside the TEM, which could be interpreted effectively by the F−N law. The results demonstrate that only 8.4% of the hemispherical cap is responsible for the electron emission. The Fe3C-filled CNT exhibit a stable field-emission behavior and can be used as an effective field emitter.



AUTHOR INFORMATION

Corresponding Author

*Fax: 86-579-82282595. E-mail: gaohuidu@zjnu.edu.cn. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by the National Science Foundation of China (No.10904129), and the Program for New Century Excellent Talents in University of Ministry of Education of China (NCET-11-1081).



REFERENCES

(1) Frank, S.; Poncharal, P.; Wang, Z. L.; De Heer, W. A. Science 1998, 280, 1744−1746. (2) Ding, L.; Zhang, Z. Y.; Liang, S. B.; Pei, T.; Wang, S.; Li, Y.; Zhou, W. W.; Liu, J.; Peng, L. M. Nat. Commun. 2012, 3, 677. (3) Hong, G. S.; Tabakman, S. M.; Welsher, W.; Wang, H. L.; Dai, H. J. J. Am. Chem. Soc. 2010, 132, 15920−15923. (4) Wan, N.; Perriat, P.; Sun, L. T.; Huang, Q. A.; Sun, J.; Xu, T. Appl. Phys. Lett. 2012, 100, 193111. (5) Chen, Q.; Wang, S.; Peng, L. M. Nanotechnology 2006, 17, 1087− 1098. (6) Cava, C. E.; Possagno, R.; Schnitzler, M. C.; Roman, P. C.; Oliveira, M. M.; Lepiensky, C. M.; Zarbin, A. J. G.; Roman, L. S. Chem. Phys. Lett. 2007, 444, 304−308. (7) Su, Q. M.; Li, J.; Zhong, G.; Du, G. H.; Xu, B. S. J. Phys. Chem. C 2011, 115, 1838−1842. 23179

dx.doi.org/10.1021/jp309121f | J. Phys. Chem. C 2012, 116, 23175−23179