Antimony on Graphite

Among various materials deposited on HOPG or amorphous carbon, particularly rich phenomena have been reported for Sb clusters (Sbn) in a size range of n = 4 to 2300 [26,76-79]. Depending on the size n of Sbn, the Sb islands formed on graphite vary from compact spheres for n = 4 to ramified fractals for n > 90. The fractal branch width decreases as n increases. These phenomena are explained in terms of the interplay of Sbn arriving rate at an existing island, and the time it takes for clusters in contact to coalesce. In all these studies, however, the possibility and consequences of Sbn decomposition were largely ignored.

In many processes, Sbn (particularly for Sb4 generated from a thermal evaporator) decomposition and/or conversion from the physisorption to chemisorption state do occur [80]. This has strong effects on compound and alloy growth involving Sb [73,74,81]. The diffusion, nucleation, and growth kinetics of chemisorbed Sb species on HOPG are expected to differ remarkably from those of physisorbed Sb4. It would be interesting to examine whether different structures form on HOPG if Sb4 decomposition is significantly activated (or similarly if the deposition flux consists of a significant percentage of Sb2 and Sbi). In addition, due to experimental difficulties, the atomic structure of Sb(111) was revealed with STM only very recently [82], and the atomic structures of other surfaces of Sb crystal are largely

Figure 6.3. (a) a-Sb lattice viewed in [111] direction (the long diagonal of rhombohedral cell). A, B, and C mark sites in three layers separated by 3.76 A. (b) Atomic configuration of graphite basal plane, a (^/3 x^3) R30° cell is outlined with dotted lines.

unknown. The crystalline Sb structures formed on HOPG allow us to image other surfaces when they appear.

We observed that spherical 3-D islands, extended 2-D islands, and 1-D crystalline nanorods are formed on HOPG. The atomic-scale STM images of the surfaces on these different-dimensional Sb structures have been obtained. Furthermore, with a low Sb4 flux and the substrate at RT, spherical islands nucleate and grow exclusively at the initial stage. In contrast, with a high Sb4 flux and substrate at ~ 100°C, formation of a spherical island is completely suppressed, resulting in only the 2-D and 1-D structures. These results are explained in terms of a relatively large difference in the activation energies of Sb4 diffusion and dissociation (or chemisorption) on HOPG, which leads to different rate changes in response to temperature increase. It is well known that the properties of different types of nanomaterials (i.e., nanoparticles, nanowires, and thin films) differ remarkably from their bulk counterparts and from each other due to their dimensionality [2,5-7,83]. Therefore, the ability to selectively grow certain dimensional nanostructural materials is highly desirable in nanoscience and nanotechnology [84,85].

The most stable a-Sb bulk crystals take a rhombohedral lattice structure [86], which can be derived by a slight distortion of a cubic lattice [87]. The distorted face-center cubic representation of a-Sb is illustrated in Figure 6.3a, with the ABC stacking at a 3.76-A layer spacing. Each site represents a base of two Sb atoms, with the other atom 5.26 A beneath the one shown in the figure. In this chapter, we use the rhombohedral index notation for the observed Sb structures [88], so Sb(0001) in Reference [82] is denoted Sb(111) here. For comparison, the atomic configuration of a graphite basal plane is sketched in Figure 6.3b.

6.4.1.1. Three Types of Sb Nanostructures

Figure 6.4a displays a STM image taken on a HOPG sample with 12 A Sb deposited at a rate of 4 A/min at RT. Three types of Sb structures are observed, labeled as 1-D, 2-D, and 3-D in the figure. The three large 3-D spherical-top islands have heights in the 50-56 nm range, and apparent lateral diameters 140-150 nm. STM

Oxygen 001 Stm

Figure 6.4. Three-dimensional view STM images of Sb structures on HOPG. (a) After 1.2-nm Sb deposited at RT and a flux of 4-A/min; (b) after 10-nm Sb deposited at RT and a 4-A/min flux. (c) and (d) are zoom-in images taken on a 2-D island. Image area: (a),(b) (1 nm)2; (c) (300 nm)2; (d) (10 nm)2. Imaging conditions: Vs = 0.45 V, It = 0.4 nA.

Figure 6.4. Three-dimensional view STM images of Sb structures on HOPG. (a) After 1.2-nm Sb deposited at RT and a flux of 4-A/min; (b) after 10-nm Sb deposited at RT and a 4-A/min flux. (c) and (d) are zoom-in images taken on a 2-D island. Image area: (a),(b) (1 nm)2; (c) (300 nm)2; (d) (10 nm)2. Imaging conditions: Vs = 0.45 V, It = 0.4 nA.

tip cannot scan underneath the nanoparticles, so we cannot know their exact shape. Based on the inertness of the substrate, the observed 3-D particles are most likely spheres or oblate spheres at this stage.

In addition to these tall 3-D islands, lower and extended structures are observed. The 2-D islands (three of them: two unlabeled ones are in the upper part of Figure 6.4a) have heights in the 3.1-3.5 nm range, and with atomically flat terraces and straight step edges. These 2-D islands are not small graphite pieces, because they grow as more Sb is deposited later. The long linear (1-D) feature in the upper part of Figure 6.4a has a height 20 nm and a measured width ~35 nm. In addition to this relatively long Sb nanorod, there are two short rods on the side of a 3-D nanoparticle in the lower half of the image. The heights of these wires are about 13 nm. It should be mentioned that these Sb rods are not necessarily formed along the steps of HOPG.

Figure 6.5. STM images of 3-D Sb islands on HOPG. (a) After 1.8-nm Sb deposited at RT and a flux of 1.8 A/min; (b) a facet on top of a 3-D island. Image area: (a) (2.5 |im)2; (b) (60 nm)2.

As more Sb is deposited in the same condition, all three types of Sb structures grow in size, but at different rates. More characteristics of these structures can be revealed as they grow. Figure 6.4b displays an image taken on a sample after 10 nm Sb was deposited. Most of the surface is covered with multilayer 2-D film domains. In the zoom-in scan in Figure 6.4c, flat terraces separated with atomic steps are observed. The measured average height of the atomic steps on the 2-D islands is 3.96 ± 0.20 A. Further zoom-in scans on a flat terrace reveal a hexagonal ordered structure as shown in Figure 6.4d, with a period 4.17 ± 0.12 A. The steps are all along the (110) direction. Comparing with the lattice parameters of Sb (111) in a-phase [86], our measured average step height is 5% larger than the bulk layer spacing (3.76 A), whereas the lateral period is shorter than the expected value (4.31 A) by 3%.

The surface of 3-D islands is smoothly curved without any facet initially. In fact, some 3-D islands of heights ~55 nm and lateral size ~200 nm still have a smoothly curved surface. This type of islands can be observed in the image of Figure 6.5a taken on a HOPG sample with 1.8 nm Sb deposited at a rate of 1.8 A/min at RT. In this condition, only 3-D spherical islands are formed at this early stage. Most 3-D islands are located along the steps on HOPG. Some islands (marked with an arrow in Figure 6.5a) have a concave top area, indicating coalescence of smaller islands to form these larger ones.

On the other hand, faceting has been observed on the surface of some relatively large 3-D islands, as illustrated in Figure 6.4b. In this image, several 3-D islands have heights in the 40-58 nm range when measured from surrounding 2-D film. Although some of these 3-D islands remain in spherical shape with a lateral dimension ~150 nm, most 3-D islands have grown to lateral size >300 nm and have developed a flat top facet with straight edges, indicating a crystalline structure inside. The top facets mostly form irregular hexagons or triangles, surrounded with a smooth curved surface. One triangular top facet is shown in Figure 6.5b. Zooming-in STM scans on such top facets reveal a hexagonal periodic atomic structure similar to that in Figure 6.4d. But the period on the 3-D top facets is 4.27 ± 0.10 A, basically the value on Sb(111) of bulk crystal. The faceted Sb crystallites observed here are similar to those observed by Kaiser et al. [79] using SEM. In their SEM image, the top hexagons of Sb islands are also surrounded with a smooth curved surface, indicating that the surrounding feature in our STM images is not totally due to tip-shape artifact.

The surface periodicity data indicate that the lattice parameters of the 3-D Sb islands on HOPG are basically the same as the bulk a-phase, with the (111) face parallel to the graphite basal plane. The lattice parameters of the 2-D films show a slight deviation from that of bulk Sb, in particular a contraction of 3% in lateral spacing. It can be seen in Figure 6.3b that the period of the (V3 x V3)R30° superstructure on graphite (0001) (a cell outlined with dotted lines in Figure 6.3b) is 4.26 A, which is quite close to the period in the (111) plane of bulk Sb. One would expect that a (111)-oriented Sb film of a few atomic layers matches with the graphite lattice underneath. But our results suggest that this is not necessarily the case. We found that the azimuthal orientation of the 2-D islands on one HOPG terrace can be different from each other, indicating that these islands do not all have a fixed azimuthal alignment with graphite substrate.

Our STM images, as those displayed in Figure 6.6, indicate that the 1-D Sb nanorods formed on HOPG are also crystalline. It is often observed that Sb nanorods grow out in two perpendicular directions, as illustrated in Figure 6.6a. Zoom-in images on Sb nanorods reveal the top surface structures mostly as rectangular periodic or rows along the rod axis. Rows displayed in Figure 6.6b, with a period of 3.70 ± 0.15 A along the rows and 4.5 ± 0.2 A between the rows are typically observed on tall nanorods with heights >20 nm, such as those shown in Figure 6.4a and those marked "T" in Figure 6.6a. Figure 6.6c shows an atomic-resolution image taken near the right-angle elbow of those relative lower Sb nanorods (marked "L", with height <15 nm). The surface lattice parameters vary from one area to another beyond experimental uncertainty. In the right-angle intersection area, we often found a nearly square order with a period of 4.18 ± 0.15 A. In contrast, a rectangular order of period (3.93 ± 0.15 A) x (4.40 ± 0.15 A) is observed away from the intersection, as shown in Figure 6.6d, with the shorter side along the axis of nanorods. A rectangular cell is outlined in Figure 6.6d, with a bright spot observable inside. The average step height on the nanorod top surface is 2.83 ± 0.20 A. As discussed in detail below, these observations suggest that these nanorods start in a simple cubic (SC) phase which forms for compressed Sb [86,89,90]. These Sb nanorods have (110) top facet, and their axes are along [110] direction.

The lattice structure of 1-D Sb nanorods deviates significantly from a-Sb bulk. For the (110) surface of bulk a-Sb, the unit cell size is 4.31 A x 4.51 A. The lattice constants measured on top of Sb nanorods indicate a compressed state. It has been observed for Group V elements such as As, Sb, and Bi that a rhombohedral to SC phase transition occurs under pressure [86,87,89-91]. For Sb, the SC phase exists in a narrow pressure range around 7.0 GPa, in which the atomic volume is about

Figure 6.6. STM images of Sb nanorods. (a) Tall (T) and low (L) nanorods growing in perpendicular directions; (b) an image on a tall nanorod showing row structure; (c) an image taken at a right-angle intersection of a low nanorod; and (d) on a low nanorod but away from the intersection. Image area: (a) (300 nm)2; (b) (10 nm)2; (c) (15 nm)2; (d) (4 nm)2. Imaging conditions: (b) Vs = 0.4 V, It = 0.3 nA; (c), (d) Vs = 0.63 V, It = 0.6 nA.

Figure 6.6. STM images of Sb nanorods. (a) Tall (T) and low (L) nanorods growing in perpendicular directions; (b) an image on a tall nanorod showing row structure; (c) an image taken at a right-angle intersection of a low nanorod; and (d) on a low nanorod but away from the intersection. Image area: (a) (300 nm)2; (b) (10 nm)2; (c) (15 nm)2; (d) (4 nm)2. Imaging conditions: (b) Vs = 0.4 V, It = 0.3 nA; (c), (d) Vs = 0.63 V, It = 0.6 nA.

85% of the normal state value and the atomic spacing is 2.966 ± 0.010 A [89]. The 4.18 ± 0.15 A square unit cells and 2.83 ± 0.20 A step height observed on the top surface at the right-angle intersections of Sb nanorods fit closely to the SC lattice, with a V2 x V2 reconstructed (001)SC surface. The nanorods grow out from the SC along the equivalent [110]SC and [110]SC directions. Away from the intersection, the nanorod lattice changes away from SC. The surface unit cell becomes rectangular, with atomic spacing expanded in the direction perpendicular to the rod axis while contracted along the axis. The transition seems continuous, with the atomic volume remaining 85% of the normal state.

Nanostructures can be in a stressed state even without an external force. For example, colloidal CdSe QDs have been observed in compressive or tensile stress state depending on how the QDs' surface is passivated [14]. The CdSe QDs are in compressive stress if the passivation layer acts as an electron donor. It is quite likely that some electrons transfer from the graphite substrate to Sb nanorods to induce the compressive stress. Similar charge transfer to the 3-D and 2-D Sb islands can also occur. Its effect on 3-D crystalline islands seems negligible due to a large volume of 3-D crystallites and a weak interaction across the interface. The 2-D films, however, are expected to have a strong interaction with the substrate, but the lattice parameter changes are fairly small. Therefore, in addition to charge transfer, the surface atomic configurations on the top and side facets (the latter cannot be imaged with STM) as well as the growth kinetics must also be considered to explain the formation and lattice structure of the Sb nanorods.

6.4.1.2. Selective Growth of Different Dimensional Sb Nanostructures

We have illustrated that Sb can grow in three different types of nanostructures on HOPG, namely 3-D clusters or crystallites, 2-D films, and 1-D nanorods. Inasmuch as different-dimensional nanostructures can have quite different properties, it is valuable to understand the mechanism behind different structural formations, and to search for conditions to selectively grow nanostructures of one type and suppress others. To this end, we investigated nanostructural formation on HOPG under different Sb flux and/or substrate temperature. The STM image shown in Figure 6.5a is taken on a HOPG sample with 1.8 nm Sb deposited at a rate of 1.8 A/min and at RT. In this condition, only 3-D spherical islands are formed at the early stage. Most 3-D islands are nucleated along the steps on HOPG. With further Sb deposition in this condition, 2-D and 1-D islands start forming.

The STM image shown in Figure 6.7 is taken on a sample with 5.4 nm Sb deposited at a rate of 18 A/min and with sample at about 100°C. Here, only

2-D and 1-D Sb structures are observed. We found that substrate temperature is more important than the flux in determining what types of Sb structures grow initially. At RT, even with a flux of 6 A/min, 3-D island nucleation and growth are dominant, whereas 3-D islands are totally suppressed with this flux but at 100°C. Raising substrate temperature further, we found that the growth cannot happen when T > 135°C. In addition, we observed that sublimation becomes significant at T > 220°C from the Sb structures grown on HOPG. After a 10-min annealing at 260°C, the nanorods almost all disappear, whereas many 2-D and 3-D islands remain. All Sb desorbs from HOPG after 10 min annealing at 375°C.

Previous studies of Sb growth on graphite have mostly addressed the nucleation and growth of compact and ramified 3-D structures [76-78,92]. Kaiser et al. [79] observed formation of other types of structures, and they obtained dominantly branched 3-D islands at a high flux (~60 A/min), whereas a few different types of structures formed at ~3 A/min. Supercooling of deposited Sb4 was considered a possible driving force for viscous fingering and dendrite crystallization. With a relative low Sb4 flux used in our experiments, however, this effect should be rather weak.

We now explain the observed phenomena in terms of a different adsorption state and diffusion rates of Sb species on HOPG. In the gas phase, Sb4 is more stable than Sb2 and Sb1. The energy cost to dissociate an Sb4 into two Sb2 is 2.4 eV (i.e., 1.2 eV per Sb2), and it is 8.4 eV for dissociating into four Sbi (i.e., 2.1 eV per atom) [93,94]. The most stable configuration of free-standing Sb4 is a tetrahedron [95]. All these can change when Sb4 is adsorbed on a substrate. For example, when deposited on Si(001) near RT, an Sb4 ball (or tetrahedron) cluster first settles in a planar dumbbell precursor chemisorption state, then dissociates into two dimers [80]. The energy barrier is ~0.7 eV for ball-to-planar transition, and is ~0.8 eV for dissociation of a planar Sb4 into two Sb2. The binding energy of Sb on Si(001) is about 0.5 eV per atom with reference to Sb4 in the gas phase, and a saturation coverage near 0.5 monolayer is maintained up to T « 800°C [93,94], much higher than the temperature at which such a coverage can be maintained on HOPG.

The binding of Sb with HOPG is expected to be significantly weaker than that with Si. Our results show that the initial sticking probability of Sb4 on HOPG is near 1 at RT, but it starts dropping noticeably at ~100°C. When an Sb4 lands on HOPG, it is most likely in a physisorption state. Similar to the case on Si(001), a physisorbed Sb4 can transfer to a chemisorption state or even dissociate into two Sb2. But these transformations require overcoming certain energy barriers, which are expected to be at least as high as those on Si(001) (i.e., >0.7 eV). The diffusion barrier Ed of physisorbed Sb4 on HOPG seems quite low (significantly less than 0.3 eV based on a comparison with nucleation and growth on some metal surfaces [96]; here we assume Ed « 0.1 eV), so that at RT they can already quickly find defect sites such as steps. Three-dimensional island nucleation occurs as several Sb4 clusters meet at a defect site. As more Sb4 clusters arrive, the existing 3-D islands grow, and more 3-D islands nucleate until the island density reaches a certain saturation value.

On the other hand, when a physisorbed Sb4 transforms to a chemisorbed Sb4 or dissociates into dimers, the diffusion barriers of these Sb species on HOPG are expected to increase significantly from that of physisorbed Sb4. We believe that the crystalline 2-D and 1-D structures are nucleated from these chemisorbed Sb species, and these nucleation events occur when the chemisorbed Sb clusters meet on a graphite terrace, not necessarily at defects. This offers a channel competing with 3-D island growth. Which channel is dominant depends on the kinetic parameters of these processes and deposition conditions.

At a relative low T (e.g., 30°C) and a low Sb4 flux, the conversion to chemisorp-tion or dissociation is strongly suppressed due to the Boltzmann factor exp[-Ec/(kT)], where Ec is the barrier of chemisorption or dissociation, and k is the Boltzmann constant. Because the diffusion barrier of physisorbed Sb4 on HOPG is quite low, migration of Sb4 clusters to step edges and other defects is highly activated. Consequently, 3-D island nucleation and growth are dominant, resulting in a sample shown in Figure 6.5.

Increasing substrate temperature enhances the rates of both Sb4 diffusion and conversion to chemisorption (or dissociation). The ratio of these two rates changes with T approximately as Rchemisorb/Rdiffusion « exp[-(Ec - Ed)/kT]. Because Ec > Ed, the increment of the chemisorption rate with T is faster than that of Sb4 diffusion, so that this ratio increases with T. Assuming Ec - Ed = 0.8 eV, this ratio at 100°C is about 300 times higher than that at RT. Correspondingly, the concentration of chemisorbed Sb species increases with T, which enhances the probability of nucleation and growth of 2-D and 1-D structures. Increasing the deposition flux further enhances this probability, because it lets chemisorbed Sb clusters more likely meet with each other.

In addition, Figures 6.4a and b show that, with a moderate flux and at RT, although all three types of Sb structures grow initially, most Sb4 clusters deposited later go to 2-D and 1-D islands. This indicates that once 2-D and 1-D islands have nucleated, they can effectively attract Sb deposited later so that the supply of Sb4 for 3-D island growth drops. At high T and high flux, this effect completely suppresses 3-D island growth, yielding only 2-D and 1-D islands on the sample as shown in Figure 6.7. At this stage, we are still searching for conditions to selectively grow either 1-D or 2-D Sb structures but not both.

Solar Power

Solar Power

Start Saving On Your Electricity Bills Using The Power of the Sun And Other Natural Resources!

Get My Free Ebook


Post a comment