Fabrication and Investigation of Nanostructures on Transition Metal

Elena Pinilla-Cienfuegos , Samuel Mañas-Valero , Efrén Navarro-Moratalla , Sergio Tatay , Alicia Forment-Aliaga , Eugenio Coronado. Applied Sciences...
0 downloads 0 Views 513KB Size
Download PDF
5334

Langmuir 2006, 22, 5334-5340

Fabrication and Investigation of Nanostructures on Transition Metal Dichalcogenide Surfaces Using a Scanning Tunneling Microscope J. B. Park,† B. Jaeckel, and B. A. Parkinson* Department of Chemistry, Colorado State UniVersity, Fort Collins, Colorado 80523 ReceiVed NoVember 21, 2005. In Final Form: March 6, 2006 Nanometer-scale holes have been fabricated on the surfaces of the semiconducting transition metal dichalcogenides (TMDCs) molybdenum ditelluride (MoTe2) and molybdenum disulfide (MoS2) by applying voltage pulses from the tip of a scanning tunneling microscope (STM) operating in ultrahigh vacuum (UHV). It was found that the tip geometry (tip shape and sharpness) influences the formation and structure of the atomic-scale nanostructures. Threshold voltage ranges for the surface modification of MoTe2 (3.0 ( 0.3 V) and MoS2 (3.4 ( 0.3 V) were determined. Negative sample voltage pulses applied to a p-type MoTe2 surface produced much larger and deeper nanometer-scale holes when compared with those produced by positive voltage pulses. The existence of threshold voltages and the pulse polarity dependence of nanostructure fabrication suggests that an electric field evaporation mechanism is applicable. Support for this mechanism was obtained by nanostructuring metallic TMDC NbSe2, where both the produced features and the threshold voltages (3.0 ( 0.3 V) were similar for both positive and negative voltage pulses.

Introduction The scanning tunneling microscope (STM) is a powerful tool that is suitable for surface nanofabrication from the nanometer scale down to the atomic scale. The STM has been used to manipulate individual atoms and molecules with atomic level precision as well as to generate artificial structures such as pits, mounds, grooves, and single atomic vacancies.1-14 These studies have been conducted using various tip materials (Pd,7 Pt,8 Al,9 Au,10 and W15,16) on substrates including graphite,8,11 Pd,7 SiO2,12 polymers,13 Ag,14 and silicon.17 Various types of nanostructures were also fabricated on transition metal dichalcogenide (TMDC) surfaces such as MoS2,18-20 WSe2,21-24 and SnSe2.18 The crystal structure of the * Corresponding author. Tel: 970-491-0504. Fax: 970-491-1801. Email: [email protected]. † Present address: Department of Chemistry and Biochemistry, University of South Carolina, Columbia, SC 29208. Tel: 803-777-6113. Fax: 803777-9521. (1) Eigler, D. M.; Schweizer, E. K. Nature 1990, 344, 524. (2) Kobayashi, A.; Grey, F.; Williams, R. S.; Aono, M. Science 1993, 259, 1724-1726. (3) Gu, Q. J.; Liu, N.; Zaho, W. B.; Ma, Z. L.; Xue, Z. Q.; Pang, S. J. Appl. Phys. Lett. 1995, 66 (14), 1747-1749. (4) Besso, K.; Hashimoto, S. Appl. Phys. Lett. 1994, 65 (17), 2142-2144. (5) Heike, S.; Hashzume, T.; Wada, Y. J. Appl. Phys. 1996, 80 (7), 41824188. (6) Heike, S.; Wada, Y.; Ashizume, T. J. Appl. Phys. 1999, 86 (8), 42204224. (7) Fukuzawa, H.; Kimijima, H.; Yoshikawa, N.; Sugahara, M. Jpn. J. Appl. Phys. 1995, 34, L1221-L1223. (8) Ma, Z.; Zhu, C.; Shen, J.; Pang, S. Vacuum 1992, 43 (11), 1115-1117. (9) Park, J. Y.; Phaneuf, R. J. J. Appl. Phys. 2002, 92 (4), 2139-2143. (10) Mascher, C.; Demaschke, B. J. Appl. Phys. 1994, 75 (10), 54385440. (11) Song, J. P.; Pryds, N. H.; Glejbol, K.; Morch, K. A.; Tholen, A. R.; Christensen, L. N. ReV. Sci. Instrum. 1993, 64 (4), 900-903. (12) Li, N.; Yoshinobu, T.; Iwasaki, H. Appl. Phys. Lett. 1999, 74 (11), 16211623. (13) Tang, S. L.; McGhie, A. J.; Lewittes, M. E. Appl. Phys. Lett. 1992, 60 (15), 1821-1823. (14) York, S. M.; Leibsle, F. M. Appl. Phys. Lett. 2001, 78 (18), 2763-2765. (15) Turchanin, A.; Freyland, W. Appl. Phys. Lett. 2005, 87, 173103. (16) Slayter, E. M. Light and Electron Microsopy; Cambridge University Press: New York, 1992. (17) Lyo, I.; Avouris, P. Science 1991, 253, 173-176. (18) Huang, J.; Sung, Y.; Leiber, C. M. Appl. Phys. Lett. 1992, 61 (13), 15281530. (19) Hosoki, S.; Hosaka, S.; Hasegawa, T. Appl. Surf. Sci. 1992, 60/61, 643647.

Figure 1. TMDCs with a crystal structure of 2H-MX2 with the important structure parameters. Top left: side view in the [110] direction; top right: view from above the (0001) van der Waals plane. The lower structural models show the different possible edge terminations for the crystal structure, which result in different numbers of DBs. Edges with a small number of DBs, such as P3 and P4, are energetically favored (also marked in the top right image).

TMDCs is shown in Figure 1. Each monolayer is formed of a layer of chalcogen atoms X, followed by a layer of the transition metal M (M ) Mo, Nb, W) and again a layer of the chalcogen atoms X (X ) S, Se, Te). Schimmel et al.22 showed that, by using STM, the application of positive voltage pulses (3.2 V) to the WSe2 surface produced small hill-shaped structures with a diameter of about 3 nm. The features were attributed to a change in the local density of states, since the atomic order was preserved in the nanostructures and their apparent heights changed at different imaging biases. In another experiment,23 voltage ramping between 1.0 and -1.0 V (20) Hosoka, S.; Hosoki, S.; Hasegawa, T.; Koyanagi, T.; Shintani, T.; Miyamoto, M. J. Vac. Sci. Technol., B 1995, 13 (6), 2813-2818. (21) Schimmel, T.; Fucus, H.; Graf, K.; Sander, R.; Lux-Steiner, M. Phys. Status Solidi 1992, 131, 89-98. (22) Schimmel, T.; Kemnitzer, R.; Kuppers, J.; Fuchs, H.; Lux-Steiner, M. Thin Solid Films 1995, 254, 147-152. (23) Garcia, R. G. Appl. Phys. Lett. 1992, 60 (16), 1960-1962. (24) Bo¨hmisch, M.; Burmeister, F.; Boneberg, J.; Leiderer, P. Appl. Phys. Lett. 1996, 69 (13), 1882-1884.

10.1021/la053148a CCC: $33.50 © 2006 American Chemical Society Published on Web 05/03/2006

Fabrication of Nanostructures on TMDCs

removed several Se atoms from the WSe2 surface, producing pits with a diameter of about 0.6 nm. Recently Jaeckel et al.25 showed that the production of defined nanostructures on WSe2 is also possible using AFM. The investigations showed that the type of structure formed is strongly dependent on humidity, tip material, pulse height and pulse length. Huang et al.18 and Hosoki et al.19 claimed that application of a positive voltage pulse to an STM tip in ultrahigh vacuum (UHV) generated single sulfur atomic vacancies on MoS2 and suggested that there is a voltage threshold for surface modification (3.6 ( 0.3 V).18 Even though nanostructure fabrication has been extensively investigated, in many cases, the exact physical process responsible for the surface modification still remains unknown. Several mechanisms such as electric field evaporation, local sublimation by tunneling electrons, local heating, local oxidation, and mechanical contact have been proposed to explain this phenomenon. The various mechanisms will be discussed below in order to clarify the possible physical processes occurring during our experiments using voltage pulses applied to MoS2, MoTe2, and NbSe2 surfaces. We concentrate on processes that can occur under UHV conditions. Therefore, the local oxidation process, which is normally observed under ambient conditions, is not discussed. It should also be kept in mind that different materials can be intrinsically very different due to the type of chemical bonding. Therefore, differences between covalently bonded (e.g., silicon), metallic-bonded (e.g., gold) or van der Waals-bonded materials (e.g., layered materials or organic molecules) can be expected. a. Electric Field Evaporation. The electric field evaporation mechanism is the basic physical process exploited in field ion microscopy (FIM), where a strong electric field applied to the tip of a sharp metallic needle ionizes He or Ne atoms and accelerates them to a phosphor screen, resulting in an image representing individual atoms on the tip.16 In the STM, when high biases are applied, a strong electric field is developed between the tip and sample. A high field can ionize and/or field-evaporate atoms from the surface of metals and semiconducting substrates. The field strength can be controlled by variation of the voltage pulse height or the tip-sample separation. Various field strengths can generate different types of nanostructures. For example, mounds and holes were generated on a Si(111) surface by adjusting the applied electric fields between the tip and sample.17 Tsong26 attempted to explain many of these experimental observations through a theoretical analysis based on an FIM model. Although the electric field evaporation mechanism has had considerable success in explaining many observations of nanostructuring, it has failed to explain several experimental results. For example, the threshold electric fields calculated for surface modification do not correspond to the experimental results. The electric fields required for the field evaporation of carbon (C f C2+) and tungsten (W f W3 +) are 85.4 and 33.7 V/nm, respectively.26 The experimental results showed that the application of voltage pulses of 10 V/nm between a W tip and graphite could remove carbon atoms from a graphite surface.27 b. Local Sublimation by Tunneling Electrons. Kondo et al.28 observed a linear relationship between the threshold voltages (Vt) for nanostructure fabrication and the bond energies of the materials, and proposed that direct electron bombardment can (25) Jaeckel, B.; Gassenbauer, Y.; Jaegermann, W.; Tomm, Y. Surf. Sci. 2005, 597 (1-3), 65-79. (26) Tsong, T. T. Phys. ReV. B 1991, 44 (24), 13703-13710. (27) Wang, C.; Bai, C.; Li, X.; Shang, G.; Lee, I.; Wang, X.; Qui, X.; Tian, F. Appl. Phys. Lett. 1996, 69 (3), 348-350. (28) Kondo, S.; Heike, S.; Lutwyche, M.; Wada, Y. J. Appl. Phys. 1995, 78 (1), 155-160.

Langmuir, Vol. 22, No. 12, 2006 5335

break chemical bonds of surface atoms, resulting in sublimation of atoms or atomic clusters. Hence, sublimation occurs only from the surface at positive sample biases, provided that the applied bias is above a threshold level, whereas negative sample biases induce sublimation of the tip materials. Even though this mechanism is in good agreement with many experimental observations, it cannot explain several other experimental observations such as the transfer of silicon clusters between a tip and sample with a positive voltage pulse,17 deposition of gold particles to the surface from a gold tip by applying a positive going voltage pulse,29 and mound formation on WSe2 using positive sample biases.22 c. Mechanical Contacts. Mechanical contact-induced nanostructure fabrication has been reported on silicon3,30,31 and gold32 surfaces. Oyabu et al.32 showed that mechanical contact itself is enough to manipulate individual Si atoms on the Si(111) 7 × 7 surface. An atomic force microscope (AFM) operated at low temperature was utilized for singe-atom manipulation, demonstrating that the manipulation processes were purely mechanical since there was no applied bias. Gu et al.3 suggested that, when a small sample bias is applied (Vbias < 0.5 V), tip sample interactions dominate, whereas, at Vbias > 0.8 V, electric field evaporation plays a key role. However, simple mechanical contact cannot explain all of the nanostructuring phenomena. For example, graphite, because of its elasticity, is not modified by mechanical contact without an applied bias.33 Koning et al.34 demonstrated that mounds can be generated on a Si(111) surface by the deposition of gold from a gold tip without mechanical contact since the gold tip was intentionally retracted from the surface during the pulse application. These results indicate that, even if mechanical contact plays an important role in nanostructure fabrication, it is not always the dominant process. d. Local Heating. The applied tunneling current and voltage from the tip can cause local resistive heating of the surface, producing nanostructures due to rearrangements of the surface atoms through migration and diffusion.35 However, one might expect a correlation between the threshold voltages for nanostructure formation and the melting or boiling points of the materials, but none is observed. However, local heating is a complex process related to phonon coupling and thermal conductivity, and so melting points and boiling points are perhaps not the only important parameters. In this work, we present experimental details of nanostructure (pit) fabrication on TMDC surfaces. We investigated three different layered materials (MoTe2, MoS2, and NbSe2) to elucidate the primary structuring mechanism under UHV conditions. Different voltage pulses (height and duration) were applied at different tip-sample separations on all surfaces. The influence of tip shape was studied in detail for MoTe2. The results strongly suggest that field emission is the primary process in the initial formation of nanostructures under UHV conditions. (29) Mamin, H. J.; Guethner, P. H.; Rugar, D. Phys. ReV. Lett. 1990, 65 (19), 2418-2421. (30) Santinacci, L.; Djenizian, T.; Schmuki, P. J. Electrochem. Soc. 2001, 148 (9), 640-646. (31) Santinacci, L.; Djenizian, T.; Schmuki, P. Appl. Phys. Lett. 2001, 79 (12), 1882-1884. (32) Oyabu, N.; Custance, Q.; Yi, I.; Sugawara, Y.; Morita, S. Phys. ReV. Lett. 2003, 90 (17), 176102. (33) Albrecht, T. R.; Dovek, M. M.; Kirk, M. D.; Lang, C. A.; Quate, C. F.; Smith, D. P. E. Appl. Phys. Lett. 1989, 55, 1727. (34) Koning, R.; Jusco, O.; Koenders, L.; Schlachetzki, A. J. Vac. Sci. Technol. B 1996, 1996 (14), 48-53. (35) Staufer, U.; Wiesendanger, R.; Eng, L.; Rothenthaler, L.; Hidber, H. R.; Guntherodt, H. J.; Garcia, N. Appl. Phys. Lett. 1987, 51 (4), 244-246.

5336 Langmuir, Vol. 22, No. 12, 2006

Park et al.

Experimental Methods Natural MoS2 and synthetic MoTe2 and NbSe2 [chemical vapor transport (I2)] crystals were utilized for studying nanostructure fabrication. All experiments were performed in a commercial Omicron multiprobe UHV system, with a base pressure of 2 × 10-10 mbar, equipped with a variable-temperature scanning tunneling microscope (VT-STM). STM measurements were performed with electrochemically etched tungsten tips. Tips were prepared from W wire (Goodfellow, 0.38 mm diameter) with a standard DC drop-off technique.36,37 This technique generates tips with apex radii ranging from 70 to 300 nm, depending on the cutoff time. Sharper tips were fabricated with a custom-built power supply that switches off the potential between the anode (W wire) and the cathode (Pt loop) within 500 ns after the lower part of the W wire drops off. The rapid removal of the voltage often results in very sharp tips with apex radii less than 30 nm. A JEOL 2000 transmission electron microscope (TEM) and a JEOL 5600 scanning electron microscope (SEM) were utilized for the characterization of the tips before and after STM experiments. Nanostructures were fabricated by interrupting the raster scanning of the STM tip during imaging, and moving the tip to a new position on the sample where a voltage pulse was applied. The tip and sample distance (gap) was reduced by 0.5 nm, with the current feedback loop being disabled before a voltage pulse was applied between the tip and the sample. Subsequently, the tip was rapidly returned to the imaging position, and the feedback loop was reactivated. After this operation, the sample surface was imaged with the same tip. The amplitude of the voltage pulses ranged between 10 and -10 V, with pulse durations from 10 µs to 70 ms. The tip-sample distance (Vsample ) 0.5 V; I ) 1 nA) during the pulses was estimated to be ∼0.6 nm from Z-spectroscopic measurements. At this tip-sample distance, the feedback loop was stabilized at a tunneling resistance of 5 × 109 Ω. All of the images were taken at room temperature in the constant current mode, with a tunneling current of 1nA at sample biases ranging from -2.0 to +2.0 V. After image acquisition, a background plane fit was applied to all of the images.

Results and Discussion Before nanostructure formation, the sample was imaged with the STM to make sure that neither natural defects nor surface deformations were present on the region of the surface to be used for nanostructure fabrication. Dopants in p-type MoTe2 and n-type MoS2 samples are imaged as protrusions and depressions at positive sample biases, respectively, whereas the contrast was reversed at negative sample biases.38 Since natural defects in the STM images are often difficult to distinguish from nanostructures (pits) produced by voltage pulses, we used positive sample imaging biases (0.1-1.5 V) for MoTe2 and negative sample biases (-0.1 to -1.0 V) for MoS2 where most natural defects are imaged as protrusions. Application of positive voltage pulses (3-10 V) produced various types of surface nanostructures on MoTe2 and MoS2 surfaces, including mounds, holes, mounds surrounding holes, and mounds inside holes. Figure 2 shows STM images of representative examples of these nanostructures produced on a MoTe2 surface. The probability of fabricating only a hole was less than 10% (see Figure 2a). In most cases, mounds were randomly distributed around the hole (see Figure 2b-d). The average depth of the fabricated holes was determined to be about 0.25 nm, which is close to the diameter of a Te atom (0.27 nm), but less than the c unit cell value of MoTe2 (0.64 nm). We then associate the holes with Te atomic vacancies. On the other hand, (36) Nakamura, Y.; Mera, Y.; Maeda, K. ReV. Sci. Instrum. 1999, 70 (8), 3373. (37) Ibe, J. P.; Bey, P. P., Jr.; Brandow, S. L.; Brizollara, R. A.; Burnham, N. A.; Diella, D. P.; Lee, K. P.; Marrian, C. R. K.; Colten, R. J. J. Vac. Sci. Technol. A 1990, 8, 3570. (38) Matthes, T. W.; Sommerhalter, C.; Rettenberger, A.; Bruker, P.; Boneberg, J.; Lux-Steiner, M.; Leiderer, P. Appl. Phys. A 1998, 66, S1007-S1011.

Figure 2. STM images (20 × 20 nm2) of the various types of nanostructures produced by voltage pulses on MoTe2: (a) a hole, (b) a mound, (c) a mound surrounding a hole, and (d) a mound inside a hole. All of the STM images were obtained at a positive sample bias (Vsample ) 0.5 V; I ) 1.0 nA).

Figure 3. Diameters (nm) of the nanostructures produced on MoTe2 as a function of the magnitude of the applied voltage pulse using two different STM tips. Applied voltage pulses for the nanostructure fabrication were between 3 and 10 V (pulse duration: 10 ms). Upper and lower bars indicate the standard deviation of nanostructure diameters at given voltage pulse conditions.

the average height of the mound-shaped features was about 0.3 nm, regardless of the applied voltage pulse heights. The height and contrast of the mound-shaped features were independent of the imaging biases, indicating that the mound-shaped features are not due to electronic effects on the tunneling current in the vicinity of the holes, but can rather be attributed to deposited material, either tellurium atoms or atomic clusters, removed from the MoTe2 surface or deposited from the tip as a result of the voltage pulses. We also investigated the influence of the voltage pulse heights on the size of the nanostructures. Figure 3 shows a plot of the average diameter of the produced nanostructures as a function of the magnitude of the applied voltage pulses obtained using two different W tips. The error bars indicate the standard deviations in the measured diameters of nanostructures. A linear relationship was obtained between the nanostructure diameters and the magnitude of voltage pulses. A threshold voltage of 3.0 ( 0.3 V for nanostructure fabrication on MoTe2 was observed, which is quite comparable to values reported for other TMDCs (MoS2: ∼3.8 V;18,28 SnSe2: ∼1.5 V;18 and WSe2: ∼2 V25,39). It was also determined that variations in the duration of the pulse between 10 µs and 70 ms did not affect the diameter of the produced nanostructures. (39) Schimmel, T.; Fuchs, H.; Akari, S.; Dransfeld, D. Appl. Phys. Lett. 1991, 58 (10), 1039-1041.

Fabrication of Nanostructures on TMDCs

Langmuir, Vol. 22, No. 12, 2006 5337

Figure 4. TEM images of STM tips with (a) 340 nm and (b) 72 nm apex radii. (c) SEM image of an STM tip with an apex radii of 25 nm.

Figure 5. Plot of nanostructure diameters (nm) produced on MoTe2 as a function of the tip apex radii. The applied pulse conditions were 3Vsample and 10 ms.

We attempted to generate smaller Te atomic vacancies using voltage pulses near the threshold value (3 V). The average diameter of holes generated by a 3 V voltage pulse varied from 1 to 20 nm, depending on which tips are used. Some tips did not produce any nanostructures, even at voltage pulses as large as 10 V, even though they were capable of atomic resolution STM images. Although, single tellurium atom detachment was not observed by applying near-threshold 1.0 ms voltage pulses. Figure 2a shows a small hole with a diameter of 3-4 nm and a depth of 0.25-0.3 nm that was produced. The area of the hole indicates that about 100 Te atoms were removed from the surface. To further investigate the effects of tip morphology, such as tip sharpness and tip shape, on nanostructure fabrication, STM tips were imaged with a TEM and SEM after using them for nanostructure fabrication. Figure 4 shows TEM and SEM images of representative STM tips. It was found that a blunt tip (Figure 4a) produced no nanostructures on the surface, whereas tips with an apex radius of less than 150 nm were able to produce surface nanostructures. Figure 4b,c shows TEM and SEM images of STM tips that reproducibly generated nanostructures and surface images where the tip apex radii were determined to be 72 and 25 nm, respectively. Figure 5 shows a plot of the diameter of created holes as a function of tip apex radius at voltages near the threshold voltage (3 V). For each tip, more than 30 nanostructures were measured to obtain good statistics for nanostructure diameter. The error bars indicate the standard deviations of the measured diameters of the holes and the tip apex radii (width of the square). The sharper tips, with apex radii less than 30 nm, were prepared by using the tip-etching circuit (see Experimental Methods section for details), whereas the tips with apex radii ranging from 70 to 150 nm were prepared without using the circuit. The linear relationship between the tip apex radii and the nanostructure diameters suggests that the size of the produced holes reflects

Figure 6. STM image (20 × 20 nm2) of a hole fabricated on MoS2. The applied pulse conditions were Vsample ) -4 V and pulse duration ) 10 ms. The image was obtained at -0.5 V and 1.0 nA. The hole diameter is ∼2.6/∼3.2 nm (min/max).

the tip apex geometry. Extrapolation of the line in Figure 5 to an artificial STM tip with a radius of 2 Å (single atom) will still produce structures with a diameter of more than 1 nm (7-12 atoms). From that conclusion, it is very unlikely that single-atom vacancies can be produced, as was claimed by Hosaka.19,20 We also attempted to fabricate similar nanostructures on an n-type MoS2 surface using negative voltage pulses. The threshold voltage for surface modification of MoS2 was determined to be 3.4 ( 0.3 V, which is close to the reported values of 3.6 ( 0.3V and ∼3.8 V by Huang et al.18 and Kondo et al.,28 respectively. Figure 6 shows an elliptical nanostructure on a MoS2 surface produced by applying voltage pulses near the threshold value (3.5 V). The major and minor axes of the hole were determined to ∼3.2 and ∼2.6 nm, respectively. Cross-sectional analysis shows that the maximum depth of the hole is about 0.25 nm, implying

5338 Langmuir, Vol. 22, No. 12, 2006

Figure 7. (a) STM image (100 × 100 nm2) (Vsample ) 1 V; I ) 1nA) of a hole fabricated by applying a negative voltage pulse (Vsample ) -5 V and pulse duration ) 0.1 ms) to a MoTe2 sample. The brighter ring around the hole is 2-3 Å higher than surrounding MoTe2 surface and has a diameter of ∼10 nm. (b) Linescan indicating the removal of complete MoTe2 layers. (c) STM image (20 × 20 nm2) (Vsample ) 0.5 V; I ) 1.3 nA) of a hexagonal-shaped hole fabricated with a negative voltage pulse (Vsample ) -3 V and pulse duration ) 0.1 ms) to the MoTe2 sample. Besides the hole, a bright rim and a less bright triangle are visible. The apparent height of the triangular feature is about 1.6 Å. (d) Linescan through the nanostructure in panel c indicating the removal of one monolayer of MoTe2.

that only first-layer sulfur atoms were removed from the surface (sulfur atom diameter: 0.32 nm). The area of the hole (about 6.6 nm2) indicates that about 80 sulfur atoms were removed from the surface. A comparison of MoTe2 (Figure 2a) and MoS2 (Figure 6) demonstrates that the structuring processes are similar for both materials. A similar feature is probably the “backwardforward” clusters, which are a kind of surface roughness, on WSe2 initiated with an AFM.25 From that we can speculate that the removal of surface atoms is possible for all semiconducting MX2 TMDCs. We also studied the application of negative voltage pulses to p-type MoTe2 to investigate whether pulse polarity affected the production of nanostructures. It was observed that negative voltage pulses produced much larger nanostructures (holes) on p-type MoTe2 when compared to positive voltage pulses. Figure 7a shows a nanostructure produced by applying a negative voltage pulse of -5 V with a duration of 10 ms. The measured diameter of the hole (15-20 nm) is about eight times greater than the average diameter of the holes (2.4 ( 0.5 nm) produced by a 5 V positive 10 ms voltage pulse. Inside the nanostructure, steps corresponding to three complete layers of MoTe2 that were removed are observed (the c-lattice constant for MoTe2 ) 0.64 nm). By applying a 3 V pulse to the sample, a hexagonal-shaped hole was fabricated, as shown in Figure 7c. A cross-sectional analysis shows that the depth of the hole is about 0.7 nm, indicating that, in this case, one complete MoTe2 layer was removed from the surface (see Figure 7d). The differences between nanostructure fabrication by positive and negative voltage pulses will be discussed later. The nanostructures produced by the negative voltage pulses (-5 V) had a bright rim surrounding the hole (see Figure 7a). The width of the bright feature was ∼10 nm and the height was less than 0.2-0.3 nm. In the case of a -3 V pulse, two bright

Park et al.

features, a smaller hexagonal-shaped rim and a more extended triangular shape, were observed around the hole (see Figure 7c). The apparent height of the triangular shaped area is about 1.6 Å. Both features are stable under continuous scanning for several hours. Similar bright edges were observed in STM images of triangular-shaped pits on WSe2 and MoS2.40-42 Several groups, including ours, have previously demonstrated that dopants in semiconducting layered materials exhibit voltage-dependent height changes when imaged with STM due to the perturbation of the local density of states.38,43,44 A sheet charge built up under an accumulation condition produced at the positive sample biases in p-type materials causes a high electrostatic force (EF) between the topmost layer of the sample and the tip apex. This leads to an increase in the measured surface height since layers weakly bound to the subsurface layers are pulled up toward the tip. Larger height changes at the step edges as a function of bias voltage were explained by additional charges localized at the edge-site dangling bond (DB) states and/or the lower work function at the edge sites due to the presence of an additional surface dipole, resulting in a stronger EF than that on the flat terrace.38,43,44 The variable brightness in the segments of the hexagonal rim around the defect in Figure 7c may be related to different edge termination geometries (see Figure 1). In principle, there are six possible morphologies of MoTe2 edge termination (see the trigonal MX2 model in Figure 1 (P1 to P6): two Te edges (P1 and P3), two Mo edges (P2 and P4), and two mixed types of Mo and Te edge termination depending on the orientation the crystal (P5 and P6). The different edge terminations have different numbers and types of DBs (P1 and P2 have four DBs, P3 and P4 have two DBs, and P5 and P6 have three DBs), resulting in variable tip sample interactions when the edges were imaged with STM. In the case of Mo and Te edges, the tip and sample interactions of the two possible edge terminations are most likely the same and thus appear as having the same brightness in the STM images due to the same number of DBs and types of atoms. The origin of the extended bright triangle in Figure 7 is currently not known. To explain the difference in nanostructure fabrication between positive and negative voltage pulses, we have to consider the electronic structure of the tunneling junction, as schematically shown in Figure 8. The application of positive sample voltage pulses to p-type MoTe2 results in the formation of an accumulation layer at the surface (see Figure 8a). This leads to sheet charges located in the first few angstroms of both the tip and sample surfaces. A sheet charge built up under an accumulation condition results in a high EF between the topmost layer of the sample and the tip apex. Because of the weak interlayer bonding in TMDCs this EF can pull up the topmost surface layer of the sample. As a result, the electric fields between the tip and sample will be spatially localized in a small area of the sample surface, and the depth of the built-in electric field will be restricted to the first few angstroms of the tellurium atoms from the topmost layer, with the result that Te atoms are field-evaporated from the surface. In contrast, negative voltage pulses cause the surface of the sample to go into depletion (see Figure 8b) or, at higher voltages, into inversion. In the depletion region, the surface charges on the tip cannot be compensated by rearrangement of the charge (40) Parkinson, B. A. J. Am. Chem. Soc. 1990, 112, 7498-7502. (41) Boneberg, J.; Lohrmann, M.; Bo¨hmisch, M.; Burmeister, M.; Lux-Steiner, M.; Leiderer, P. Z. Phys. B: Condens. Matter 1996, 99, 567-570. (42) Akari, S.; Moller, R.; Dransfeld, D. Appl. Phys. Lett. 1991, 59 (2), 243245. (43) Schlaf, R.; Louder, D.; Nelson, M. W.; Parkinson, B. A. J. Vac. Sci. Technol. A 1997, 15, 1466-1472. (44) Schlaf, R.; Schroeder, P. G.; Nelson, M. W.; Stubner, R.; Tiefenbacher, S.; Jungblut, H.; Parkinson, B. A. Thin Solid Films 1998, 331, 203-209.

Fabrication of Nanostructures on TMDCs

Langmuir, Vol. 22, No. 12, 2006 5339

Figure 9. Nanostructuring voltage dependence for NbSe2. A symmetric behavior is observed for both positive and negative polarity of the applied voltages pulses. Type 1 corresponds to hole formation, Type 2 to cluster formation, and Type 3 to crater-like structures.

Figure 8. (a) Positive sample voltage pulse causing hole accumulation in the p-type semiconductor, resulting in a strong (a few angstroms) electric field between the topmost layer and the tip apex, which results in the detachment of only top surface Te atoms. (b) Negative sample voltage pulse causing hole depletion in the p-type semiconductor. The hole depletion (or inversion) leads to electric field penetration deeper into the sample, resulting in the detachment of several layers. The opposite is applicable to n-MoSe2 nanostructuring.

densities within the top layer of the sample. Since there are very few carriers (electrons), the field penetrates into the sample surface, creating a relatively thick space charge layer. This means that the EF between the tip and the surface layer of the sample is weaker, resulting in a much weaker force on the top layer than that under accumulation. Therefore, more than one complete layer will be effected under the influence of the built-in electric fields, and several layers can be removed from the sample with a voltage pulse. The electric field evaporation model described above would predict that nanostructuring similar to that observed under accumulation would occur with either positive or negative voltage pulses on a metallic sample. NbSe2 is a metallic TMDC, and so we investigated the nanostructures produced by different biases on this material. The experimental results on NbSe2 show a symmetric behavior for nanostructure creation for pulses of both polarities (see Figure 9). A threshold voltage of (3 V was determined, which is in the same range as the threshold voltages for MoS2 and MoTe2. Above the threshold voltage, three types of nanostructures (holes (Type 1), mounds (Type 2), and craterlike structures (Type 3)) were observed. Figure 9 shows a plot of the percentage of the different types of structures versus the applied voltage. At voltages just above the threshold voltage, mostly cluster-like structures appear, whereas, at higher voltages, the crater-like structures are favored. The plot displays a similar behavior for both negative and positive sample biases for producing all three types of structures. A space charge region behavior, such as that in the semiconducting TMDCs MoS2 and MoTe2, is not possible because NbSe2 is metallic. Therefore, for the nanostructuring of NbSe2, the model in Figure 8a is valid for

both polarities and supports the model of the influence of a space charge region for the nanostructuring process. Now that we have established the role of the field evaporation mechanism, we can discuss whether any of the other mechanisms outlined in the Introduction have a role in the voltage pulse nanostructuring of TMDCs. We can rule out a mechanical contact mechanism for the initial formation of nanostructures on NbSe2, MoTe2, and MoS2 because Fuchs et al.45 showed that soft STMtip contact only produces small plastic surface deformations on WSe2. It cannot be excluded completely that, after the initial formation of the nanostructure, mechanical contact plays a role. TMDC surfaces are soft and pliant, like graphite, and so tend to recover from most mechanical contacts with the tip. The generation of the larger and deeper holes with negative voltage pulses is opposite to what is expected from the electron bombardment mechanism, as suggested by Kondo et al.,28 where negative voltage pulses cause only the tip materials to be etched, with no damage to the substrates. The existence of threshold voltages for the surface modification and the pulse polarity dependence of nanostructure fabrication suggest that the electric field evaporation mechanism is applicable to nanostructure fabrication in MoTe2 and MoS2. This is supported by the fact that voltage pulses applied without moving the tip toward the surface do not generate nanostructures. The application of larger voltage pulses leads primarily to an increase in structure dimension, as shown in Figure 3. However, even though field evaporation may be the mechanism through which the modification process is initiated, other processes such as thermal and/or current effects may contribute to the later stages of material removal.

Conclusions We have used an STM to generate atomic-scale nanostructures on semiconducting MoTe2 and MoS2 as well as on metallic NbSe2 surfaces by applying voltage pulses between the tip and the sample. Similar distinct threshold voltages for the surface modification of MoTe2 (3.0 ( 0.3V), MoS2 (3.4 ( 0.3V), and NbSe2 (( 3.0 ( 0.3V) surfaces were measured. It was found that the geometry of the STM tips (tip sharpness and shape) plays an important role in nanostructure fabrication on MoS2 and MoTe2. An extrapolation of these data predicts that very small tip apex radii could produce atomic-scale features. Positive voltage (45) Fuchs, H.; Laschinski, R.; Schimmel, T. Europhys. Lett. 1990, 13 (4), 307-311.

5340 Langmuir, Vol. 22, No. 12, 2006

pulses applied to the p-type sample removed only chalcogens (Te or S atoms) from the topmost surface layer, whereas negative voltage pulses etched away more than one complete layer of MoTe2. This polarity dependence of nanostructure fabrication was explained by the difference in the magnitude and penetration depth of the applied electric fields. The model is supported by the experiments on the metallic TMDC, NbSe2, for which

Park et al.

similar nanostructuring with either positive or negative voltages occurs. Acknowledgment. The authors thank Paul Schroeder and Brian France for experimental assistance and acknowledge the financial support of the Petroleum Research Fund of the American Chemical Society under contract # PRF 34383-ACS. LA053148A