Unidirectional lateral nanowire formation during the epitaxial growth of GaAsBi on vicinal substrates

We report on enhanced control of the growth of lateral GaAs nanowires (NWs) embedded in epitaxial (100) GaAsBi thin films enabled by the use of vicinal substrates and the growth-condition dependent role of Bi as a surfactant. Enhanced step-flow growth is achieved through the use of vicinal substrates and yields unidirectional nanowire growth. The addition of Bi during GaAsBi growth enhances Ga adatom diffusion anisotropy and modifies incorporation rates at steps in comparison to GaAs growth yielding lower density but longer NWs. The NWs grown on vicinal substrates grew unidirectionally towards the misorientation direction when Bi was present. The III/V flux ratio significantly impacts the size, shape and density of the resulting NWs. These results suggest that utilizing growth conditions which enhance step-flow growth enable enhanced control of lateral nanostructures.

(Some figures may appear in colour only in the online journal) Nanowires (NWs) are a cornerstone of next generation electronic and optical devices, providing the unique opportunity to employ bottom-up self-assembly to create novel semiconductor nanostructures [1,2]. NWs boast excellent material quality thanks to their ease of fabrication utilizing the vapor-liquidsolid (VLS) growth mode generating well-defined geometries [3,4]. In comparison to vertical NWs, planar NWs have the added attraction of being compatible with traditional planar processing [3,5]. The VLS growth mode utilizes a metallic nanoparticle; in the III-V compound semiconductor materials system the nanoparticle is most often Au. Au has proven to be an effective nanoparticle in the III-V system, yielding a large body of work utilizing Au nanoparticle pre-deposition as a seed, or template layer, for the VLS growth of both planar and vertical NWs [3,[5][6][7][8]. However, Au is known to be a contaminant in III-V VLS grown NWs [9,10]. Therefore, the use of nanoparticle-free or self-catalyzing VLS NW growth is of significant interest [11,12]. III-V NWs most commonly grow perpendicular on (111)B oriented substrates due to its low surface energy [13,14]. For lateral NWs the preference is to grow along the surface projections of the 〈111〉B crystallographic direction [3]. Thus, the growth direction is fixed by the low surface energy of the (111)B plane, however, NWs grow in the+and − 〈111〉B direction with relatively equal probabilities [3,13,15]. Unidirectional lateral NWs have only been realized by limiting the surface projections of the 〈111〉B direction [16]. Additionally, Ga nanoparticle self-propulsion was observed to be unidirectional by inherent wafer miscut morphology [17]. Despite the large body of work on NW synthesis, the preferential steering of lateral NWs remains relatively unexplored.
In the current study, we demonstrate unidirectional lateral Ga-catalyzed NW growth which are self-forming during Garich GaAsBi epitaxy on GaAs. The NWs are embedded in the epitaxial film due to the concurrent 〈100〉 growth of the GaAsBi film and the surface Ga nanoparticle forms lateral NWs. Embedded NWs have the potential to enhance performance due the inherent in situ surface passivation and epitaxial interfaces. As a result of the concurrent growth and preferential VLS growth of GaAs at the liquid nanoparticle-solid interface, the embedded NWs appear to be GaAs, i.e. Bi incorporation is excluded in the NW, see figure 1 [18]. Thus far, only Steele et al and Wood et al have published on embedded NWs [18,19]. The realization of unidirectional NWs is important and we identify step-flow growth facilitated by the use of a miscut substrate and the behavior of Bi as a surfactant as key driving forces controlling NW growth directionality during epitaxy.
Step density and direction can be controlled through the substrate miscut prior to growth. However, the step morphology can be modified during growth by growth conditions and by the addition of impurities and surfactants, which modify adatom migration and preferential incorporation sites and incorporation rates [20][21][22][23][24][25]. Limited steering of NWs has been achieved utilizing substrate orientation, pre-growth annealing conditions, polarity or chemical composition of the NW [6,16,26,27], but the full parameter space that controls their movement is not well characterized. In this letter, we demonstrate control of the NW growth direction and density by utilizing vicinal surfaces to select the growth direction and the addition of Bi which segregates at the surface modifying growth dynamics to vary the shape, density, and unidirectionality of lateral NW growth.
The samples in this study were synthesized using a Riber 2300 molecular beam epitaxy (MBE) system with a stationary substrate, i.e. non-rotating, growth scheme. For the three GaAsBi samples, a 500 nm GaAs buffer is first grown on the substrate at 580°C, followed by 0.4 monolayers of Bi predeposition. After this step, 250 nm of GaAsBi is grown at 320°C. For the GaAs samples, the same buffer layer is grown, and then the temperature is reduced to 320°C for the growth of the subsequent 250 nm GaAs layer. The stationary growth condition enables the III/V flux ratio to smoothly vary across the sample. In our case, the spatial variation of growth conditions is from As-rich growth to Ga-rich growth [28]. During Ga-rich growth, Ga accumulates on the surface and we can discern the impact the III/V flux ratio on Ga nanoparticle, size, shape, density and III/V flux ratio droplet onset. The relationship between flux ratio and Bi incorporation in GaAsBi is well understood [29]. Our prior work investigated the use of vicinal substrates on Bi incorporation [28]. Wood et al studied GaAsBi growth under Ga-rich conditions on vicinal substrates oriented in the 〈111〉A direction (GaAsBi A-step sample) with embedded lateral NWs that are primarily GaAs in a GaAsBi matrix [18]. The Bi incorporates into the Ga nanoparticles but appears to develop as inclusions and not participate directly in the GaAs VLS growth creating the NWs [18]. The tunneling electron microscopy (TEM) image of a NW in the GaAsBi A-step sample is imaged in figure 1, this image is reproduced from Wood, et al's work [18]. In the wake of the GaAsBi and GaAs lateral NW growth we observe raised trails (greater thickness) using atomic force microscopy (AFM) and infer that the NW growth is faster than the concurrent epitaxial film growth.
To characterize the direction, size, shape, and density of the NWs, we used a Digital Instruments Dimension 3100 AFM and Nanoscope Analysis software. All measurements in this paper are by AFM. We characterized the lateral NW using the characteristics of the raised trail of film in the wake of the moving Ga nanoparticle. The extruded section observed by AFM is a product of the increased growth rate due to the NW growth embedded within the 〈100〉 oriented GaAsBi film. Both TEM and AFM measurements indicate that the trail length is ∼4 μm for nanoparticles nucleating spatially where the III/V flux ratio is ∼1-1.4. This indicates that the AFM measurements, despite sensing only surface characteristics, is sensitive to the entire embedded NW.
At the low temperature required for GaAsBi growth, we expect GaAs growth to occur in a three-dimensional growth mode [30]. Two behaviors that modify this that are at play in our samples are: (1) the reentrant behavior of mound formation with increasing As 2 flux and (2) the impact of Bi which acts as a surfactant on the surface and incorporates into the film [31][32][33]. We introduce in these experiments an additional factor modifying the growth mode, which is the use of misoriented substrates to increase the linear step density in either the [110] or [1][2][3][4][5][6][7][8][9][10] direction. Herein, steps due to the miscut from (001) to the [110] direction (Ga terminated, (111)A face) will be called A-steps and steps due to the miscut from the (001) to the [1-10] direction (As terminated, (111)B face) will be called B-steps.
Vicinal surfaces enhance step-flow growth by reducing the distance a Ga adatom must diffuse before incorporating into a step edge, driving the balance between step incorporation versus island nucleation towards that of step incorporation [34,35]. Incorporation and morphology evolution are significantly modified by the anisotropies introduced by the different reactivity of the A-steps and B-steps, particularly the anisotropic diffusion of Ga [35]. The B-step has a larger step velocity than the A-steps for MBE grown GaAs [20,31,36]. Surfactants, however, significantly impact growth dynamics including direction-dependent adatom diffusion. For example, Wixom et al found that Bi as a surfactant in the GaAs system enhances the lateral growth rate in the [110] by almost 300% and had negligible effects on the [1-10] lateral growth rate [37]. The low growth temperatures and step edge kinetics move the system away from equilibrium, while the misorientation increases the large number of reaction sites driving the system towards equilibrium. Hence, the interplay between temperature and step density provide a unique platform for controlling NW growth.
This study reveals the impact of the presence of Bi and substrate misorientation on lateral NW growth. Table 1 outlines the six distinct growth conditions for the six samples. GaAs and GaAsBi were each grown on GaAs substrates with three different surfaces: (001), (001) vicinal misoriented toward the (111)A by 3°, and (001) vicinal misoriented toward the (111)B by 3°. A 3°miscut yields an approximate terrace length of 5.4 nm, assuming a surface atomic bilayer. The first column in table 1 outlines the names of the samples from which they will be referred to within this text. The NW direction column in table 1 describes along which plane the NWs are aligned. Also, further outlined is whether the NWs are unidirectional or bidirectional. Undulations created during growth in the Ga-rich regime are in the [111]B direction, except for the GaAsBi B-step sample wherein no discernable undulations were found, consistent with literature on low temperature GaAs growth [34]. The last column in the table is the amplitude of the surface undulations; the amplitude results from the limited diffusion of Ga on the surface resulting in step bunching and therefore surface mounds stretching long in the [1-10] direction. The GaAsBi (singular) sample has the largest undulation amplitude, suggesting strongly enhanced anisotropy in the Ga diffusion length in the [110] and [1][2][3][4][5][6][7][8][9][10] directions.
Typically, NW synthesis through a VLS mechanism exploits pre-deposited metal droplets (nanoparticles) on a substrate which are exposed to the NW constituent materials [1]. When supersaturation of the constituent material(s) is reached, nucleation of the NW occurs at the nanoparticle interface with the solid crystal. The NW formation continues so long as growth materials are provided [1]. The nanoparticles in this experiment are not pre-deposited, but rather form due to excess Ga on the surface during epitaxy. The nanoparticles herein are self-catalyzing, meaning that NW growth consumes the nanoparticle material. This experiment studies two simultaneous processes that have been previously studied independently: (1) the formation of Ga nanoparticles during epitaxy due to excess Ga on the surface determined by III/V flux ratio, the behavior of Bi as a surfactant, and the introduction of controlled step density and direction, and (2) Ga-catalyzed VLS NW growth. The formation of Ga nanoparticles is determined by excess Ga on the surface with respect to the Group V constituents, in this case, As and Bi, during epitaxy. The excess Ga spatially varies across the sample in relation to the flux distributions. In addition, two factors modify surface dynamics and, therefore, the incorporation rate of Ga and growth of GaAsBi. Under these Note. * indicates that the sample was mostly unidirectional, but ∼8% of NW growth is in the opposite direction.
growth conditions Bi both incorporates and segregates at the surface: the III/V flux ratio modifies the ratio of incorporated to segregated Bi which, in turn, modifies the Ga incorporation rate. Furthermore, the use of vicinal substrates modifies the Ga incorporation rate due to the different chemistry and binding of Ga to A-and B-steps. The embedded lateral NW is particularly interesting and has only recently been observed as a consequence of studies exploring the incorporation of Bi into GaAs which necessitates Ga-rich growth conditions. Figures 2(a)-(d) is a schematic of the lateral embedded NW growth mechanism sketched with the blue trail attached to the nanoparticle above the epitaxial film, representing the trail which can be measured using AFM. The bottom row, figure 2(e), shows AFM images of the GaAsBi A-step sample as a function of increasing III/V flux ratio. In the As-rich regime, undulations on the surface are observed associated with anisotropic Ga diffusion length and step-bunching, and in the Ga-rich regime nanoparticles form which increase in density with III/V flux ratio. What distinguishes these NWs from most other NW studies is that 〈100〉 film growth is occurring simultaneously with lateral NW growth. During the VLS growth of NWs using self-catalyzing nanoparticles, a dynamic balance between the consumption and accumulation of the group III species must be achieved [11]. Self-catalyzed growth depends critically on growth temperature and III/V flux ratio. Generally, III-V NWs grow in the 〈111〉B direction. In the case of lateral NWs on (100) substrates growth is along the surface projections of the 〈111〉 B direction [3]. Epitaxial growth of NWs on GaAs(100) substrates is out-of-plane in the 〈111〉B direction, with equal probability of growth in either of the two 〈111〉B directions [3,13,16]. This implies that for lateral growth on a (100) substrate, there is equal probability of growth in the [0 − 1 1] or [0 1 −1] direction, yielding bidirectional, but aligned NWs. The 〈111〉B direction is preferred because it minimizes the surface energy at the liquid-solid interface where nucleation occurs [38,39]. However, depending on the nanoparticle and substrate, lateral NWs moving in the 〈111〉B and the 〈111〉A direction have been observed [5,6,15,18,19,40]. Reports with In, Au and Bi nanoparticles report lateral NW growth via VLS with growth in the 〈111〉B direction regularly explained using the low surface energy of the plane guiding the growth direction [5,15,19,40]. Alternatively, observed NW growth in the 〈111〉A direction with Au nanoparticles is argued as a maximization of contact with the low surface energy plane [6,27,41]. This literature suggests that with careful architecture of the relative surface energies at the nanoparticlesolid interface, there is great potential in preferentially steering NWs along desired crystallographic directions.
The two GaAs samples have NWs that move along the same crystallographic direction, however achieving bidirectional versus unidirectional movement is dependent on the direction of the miscut. Figure 2 shows AFM images of nanoparticles and the extruded NW trails formed on the surface. Looking first at GaAs growth (figures 3(a) and (b)), we see that the nanoparticles terminate triangular-shaped trails aligned in the ± [1-10] direction. While the use of a vicinal substrates will generally enhance step-flow growth in the direction of the step motion, the low temperature used in these studies mitigate this effect, especially for GaAs growth which takes place without surfactant action, which we expect for GaAsBi. The nanoparticle motion we observe is in the same direction for both vicinal GaAs samples, independent of the miscut direction. Therefore, the modification of Ga dynamics and migration lengths, due to the different chemistry and incorporation rates of Ga at the A-and B-steps, are not  [42]. However, this difference is clearly not large enough to impact VLS nucleation without surfactant action. When the surface energy is driving the NW growth, there is equal probability of traveling in the + or − [1-10] direction [3]. The GaAs A-step sample growth is bidirectional, however the GaAs B-step sample is largely unidirectional implying the step-flow growth velocity is enhanced by adding B-steps and drives the direction of the NW growth. The anisotropy in this behavior is consistent with the fact that Ga migration is known to be anisotropic and enhanced in the B direction due to the anisotropy in the incorporation rates at the two steps. In the [111]B direction, the step-flow propelled nanoparticles to move in a unidirectional manor aligned with the low surface energy direction 〈111〉B. We should point out that approximately 8% of nanoparticles move in the opposite direction. This behavior is consistent with the presence of domains with different growth modes.
With the addition of Bi during growth, we observe significant differences in the behavior of the NW growth and nanoparticle characteristics. The GaAsBi vicinal samples found in figures 3(c) and (d). The addition of Bi during GaAs growth decreases the density and increases the size of Ga nanoparticles, consistent with an increased Ga diffusion length. Unlike GaAs nanoparticle/trail motion, both A and B misoriented substrates have unidirectional nanoparticle movement. We can determine that the addition of Bi enhances step-flow growth velocity in both [111]A and [111]B directions. Therefore, we conclude that the smaller difference in (111)A and (111)B interface energy at lower temperature, plus the enhancement of step-flow growth velocity with the addition of Bi is significant enough to drive NW growth in a controlled fashion in both 〈111〉A and 〈111〉B directions. We speculate that the addition of Bi may also modify the interfacial energies. Bi is not only on the surface as a segregant/ surfactant, but also in the nanoparticle [18], resulting in variations to both the surface energies and the chemical potential of the nanoparticle. In summary, while the GaAs A-step sample demonstrated bidirectional NW growth, unidirectional growth was achieved through the use of vicinal substrates and the addition of Bi. The enhancement of step-flow growth velocity through the addition of Bi also enabled a 90°change in direction of the trails/nanoparticle motion for the GaAsBi A-step sample. Analysis of the amount of excess Ga on the surface and how it is modified by the addition of steps and Bi is useful towards developing an understanding of the incorporation of Bi in GaAs. The density of Ga nanoparticles as a function of position and, therefore, III/V flux ratio, is shown in figure 4. GaAsBi growth introduces delayed onset of Ga nanoparticle formation in comparison to GaAs (delayed to a III/V flux ratio〉1), which is evidence of surfactant-like effects on Ga incorporation from Bi segregation. We speculate that Bi introduces enhanced step velocity and an increased Ga incorporation delaying excess Ga buildup on the surface. Numerous studies have shown that Bi incorporates into GaAs only under Ga-rich growth conditions (III/V flux ratio 〉1), and the Bi incorporation profiles, determined by x-ray diffraction, for the GaAsBi samples studied herein are published elsewhere [28]. Therefore, in the As-rich growth regime Bi segregates at the surface. GaAs growth on the miscut substrates exhibit nanoparticle formation at a III/V flux ratio of ∼0.8, shown in figure 4. In comparison, GaAs growth on a singular (100) substrate manifests nanoparticle formation at a flux ratio of ∼0.9. All the samples with Bi addition show a delayed nanoparticle onset to a III/V flux ratio of ∼1.04.
In addition to the higher III/V flux ratio needed for excess Ga accumulation on the surface, figure 4 shows that the addition of Bi causes a decrease in nanoparticle density. GaAs surfaces have a Ga nanoparticle density roughly 13 times that of the GaAsBi surfaces. The average GaAs nanoparticle density is 2.2 nanoparticles μm −2 and the maximum density of nanoparticles on the GaAsBi surface is 0.17 nanoparticles μm −2 . The density of nanoparticles is indicative of the Ga diffusion length on the surface, and Bi is known to increase the Ga diffusion rate on GaAs when acting as a surfactant, decreasing the density [37]. As expected, both GaAs and GaAsBi films show an increase in nanoparticle density with increasing Ga/As flux ratio.
Both the GaAs and GaAsBi A-step show a steady increase in nanoparticle density with increasing III/V ratio. On the other hand, the Ga nanoparticle density on the GaAsBi B-step saturates at a 1.15 flux ratio which may be due to a saturation of Bi incorporation at a flux ratio of 1.05 [28]. The GaAsBi B-step sample incorporates the least amount of Bi into the film (2.0%, 2.2% and 2.4% for GaAsBi B-step, singular, and A-step, respectively), implying that this sample has the largest amount of excess Bi on the surface to act as a surfactant. The increase in the nanoparticle density, plotted in figure 4, is consistent with a decrease in Ga diffusion length and with the reduction of Bi on the surface due to increased incorporation into the film, ultimately reducing surfactant impact on growth dynamics. As such, the relative nanoparticle densities correlate with Bi incorporation into the GaAsBi film. For example, GaAsBi A-step incorporated the most Bi into the film, leaving the least amount of Bi on the surface (compared to the other two samples), yielding a decreased surface diffusion rate, which in turn increases the nanoparticle density. Overall, we infer that the introduction of Bi on the surface significantly and differentially (for A-and B-steps) changes the incorporation rate of Ga during GaAs/ GaAsBi growth.
In figure 4 the GaAs A-step sample does not act like the other samples in the As-rich regime. There is a much higher density of nanoparticles in the As-rich regime for the GaAs A-step sample, which decreases with increasing III/V flux ratio and settles to roughly the same density near stoichiometric conditions. To explain this, we must consider that over the flux ratio range investigated the B-step surface incorporates Ga more efficiently than the A-step surface [25]. The nanoparticles in the As-rich regime for the GaAs A-step sample are aligned along well-defined mounds on the surface in the [1-10] direction. These mounds arise due to the Erlich-Schwoebel instability from the differences in down and up step crossing probabilities of Ga. In the Ga-rich regime the behavior of the GaAs A-step and B-step samples is similar, indicating that step crossing anisotropy becomes insignificant.
As a means of quantifying the excess Ga volume, we utilized the AFM bearing depth analysis tool in the Nanoscope analysis software. Figure 5 shows the total excess Ga approximated by the total Ga nanoparticle volume in a 5×5 μm AFM image plotted as a function of III/V flux ratio. The excess Ga is approximated by defining a plane that selects only the volume of the nanoparticle, this plane selection was above the NW trails, so some of the Ga volume (that etched into the film and the disc-shaped cross-section of height consistent with the NW trail) is not included in this analysis. The three GaAs samples show a steady increase in excess Ga with increasing Ga flux up until a flux ratio of 1.10, where the excess may plateau. Consistent with the earlier discussion of nanoparticle III/V flux ratio onset, the addition of Bi shows that Bi significantly decreases the excess Ga (compared to the GaAs samples), due to an increase in incorporation rate. It should be noted that the nanoparticle volume that we are defining as excess Ga, will also include any excess Bi that the nanoparticle absorbs, but we expect most of the droplet to be Ga (observed by TEM for the GaAsBi A-step sample [18]). The GaAsBi singular and A-step samples exhibit a steady increase in excess Ga with increasing Ga-flux. The GaAsBi B-step sample at first tracks The length of the NWs, as inferred from the trail length, is increased through the addition of Bi. NW length as a function of III/V ratio is plotted in figure 6(a). Bi increases the length of the NW by approximately 6-fold, this additional velocity is believed to be derived from the enhanced step-flow velocity facilitating NW growth driving nanoparticle motion. Considering the length of the NWs to be directly correlated with the nanoparticle velocity, we can conclude that the GaAsBi B-step experiences the largest step-flow enhancement due to Bi but that it is flux ratio dependent. The change of trail length with III/V flux ratio is small for the GaAsBi A-step   sample and large for the GaAsBi B-step sample. This is consistent with Miwa and Nishinga's studies of the dependence of Ga incorporation at A-and B-steps on flux ratio [43]. They found that the Ga incorporation at A-steps is independent of III/V ratio while that at B-steps decreases with increasing III/V flux ratio [43]. Knowing the growth times for the GaAs/GaAsBi epitaxial films, we can calculate the approximate growth rates of the NWs. The growth rates calculated from the length of the trails are 0.33 nm s −1 for GaAs, 1.7-2.5 nm s −1 and 1.3-2.2 nm s −1 for GaAsBi A-step and B-step samples respectively. The Ga nanoparticle diameter reflects similar trends to the NW length as shown in figure 6(b). This validates that the Ga diffusion rate determines the size and density of nanoparticles and plays an integral role in determining the velocity of the NW growth. This is consistent with literature on step-flow growth for vicinal GaAs with Bi enhancing this anisotropy [37,43].
Schematic diagrams of the trail shapes are shown in the last column of figure 3, demonstrating clear faceting above the film surface from the lenticular NW growing beneath the surface [18] (schematic also in figure 2(d) for GaAsBi A-step sample). Cross sectional data on the GaAsBi A-step and B-step samples are taken approximately ∼3 μm from the trail end (∼2/3 of the way up the NW). This location was chosen because it is far enough away from the nanoparticle to isolate the measurement from any post growth effects from the nanoparticle, but far enough into the trail that the trail was well developed and relatively large. For the GaAs samples, the measurement was taken at the approximate center of the trails, due to the limited length of the trails. The cross-section measurements highlight the differences in the different step character and its role in defining the NW shape. These differences arise from the anisotropic lateral growth rates in GaAs and are attributed to a combination of different sticking coefficients at the [−110] and [110] steps along with the asymmetry for the different surface reconstructions varying the diffusions coefficients [37].
Three characteristic measurements of the cross sections of the NW trails are shown in figure 7: (a) height, (b) length of the base (which reflects the NW diameter) and (c) length of the top facet (runs parallel to the base). Examination of the cross sections of the NW shows that Bi increases the diameter of the NW, consistent with the change in nanoparticle size. The GaAsBi A-step sample is more triangular and never fully develops a top flat facet, and therefore it was not characterized as having a top facet in figure 7(c). The GaAsBi B-step sample experiences the most drastic increase in size and this increase is observed in all three characteristic measurements. Three main conclusions can be drawn: (1) the samples with Bi experience an increase in width, (2) the flat plateau top of the NW is wider with Bi and (3) only the GaAsBi B-step sample experiences an enhanced out-of-plane growth rate creating NW trails that were almost twice as high as the other samples. It is interesting to note that despite the GaAsBi B-step sample having on average larger values in all three measures, the aspect ratio of these parameters for the GaAsBi A-step and GaAsBi B-step (data not shown) remain very similar indicating that similar surface energies control the shape of the NW when Bi is in the film, despite the different miscut substrates.
The trails are triangular when viewed from above, which implies that the nanoparticle diameter is increasing as the NW forms throughout the epitaxial film growth. Correlations between the nanoparticle diameter, NW length and NW diameter (base) can be found in figure 8. The nanoparticle diameter linearly scales with the trail length and the base of the NW for the GaAsBi B-step sample. All other samples have minimal variation or no correlation. It is possible that the GaAsBi A-step sample reaches a maximum nanoparticle diameter and then seeds more nanoparticles with excess Ga content, yielding little variation in NW length and width. It was expected that the trail length is proportional to the step velocity, and only the GaAsBi B-step shows an increase in length and lateral growth rate with a linear relationship. In figure 8(b), the length of the base is plotted against the NW length; effectively this is the lateral growth rate versus the velocity of the nanoparticle. Only the GaAsBi B-step sample showed a dependence between the two, where the longer NW yielding a wider NW. The dependence on III/V flux ratio of the trail length and nanoparticle diameter (figure 6), along with the characteristics in figure 8, demonstrate that the linear relationship is likely due to the differences in the nature of the different step character between the two samples. The GaAsBi B-step sample has both A-steps and B-steps present on the surface where the B-steps are generating the dependence on III/V ratio [43,44]. Whereas, the GaAsBi A-step sample can be characterized as having only A-steps, so the NW growth is III/V flux ratio independent [43,44]. There was little variation found between the GaAs vicinal samples, but this anisotropy (but to a lesser degree) is expected to be present. This observation of anisotropy may be damped due to the oxide growth, the NWs being buried, and the smaller sizes of the NWs in the GaAs samples. The presence Bi amplifies the differences in the step character, making trends are more annunciated in the GaAsBi vicinal samples.
In conclusion, we use a surfactant (Bi) and vicinal surfaces to enhance the control of lateral growth of GaAs NWs embedded in epitaxial GaAsBi films. The presence of Bi increases the Ga diffusion and incorporation rates yielding larger unidirectional NWs than in GaAs films, but also delays NW onset to higher III/V ratios than were required for GaAs films. The unidirectional growth was guided by the misorientation of the substrate for the GaAsBi films: [111]A or [111]B. Without Bi, the NWs follow the lowest surface energy facet, [111]B. We also use the III/V flux ratio to vary the size, shape, and density of the nanoparticles.