Growth of Self-Catalyzed InAs/InSb Axial Heterostructured Nanowires: Experiment and Theory

The growth mechanisms of self-catalyzed InAs/InSb axial nanowire heterostructures are thoroughly investigated as a function of the In and Sb line pressures and growth time. Some interesting phenomena are observed and analyzed. In particular, the presence of In droplet on top of InSb segment is shown to be essential for forming axial heterostructures in the self-catalyzed vapor-liquid-solid mode. Axial versus radial growth rates of InSb segment are investigated under different growth conditions and described within a dedicated model containing no free parameters. It is shown that widening of InSb segment with respect to InAs stem is controlled by the vapor-solid growth on the nanowire sidewalls rather than by the droplet swelling. The In droplet can even shrink smaller than the nanowire facet under Sb-rich conditions. These results shed more light on the growth mechanisms of self-catalyzed heterostructures and give clear route for engineering the morphology of InAs/InSb axial nanowire heterostructures for different applications.


S1. Measured parameters of the NWs of all sample series
In order to measure the nanoparticle (NP) height H and base radius Rd, the InSb segment length L and maximum diameter D, the nanowires (NWs) were mechanically transferred from as grown substrates onto Si substrates and 90° projection images were taken with scanning electron microscopy (SEM). We measured about 30 NWs for each sample and we calculated the NP aspect ratio (H/Rd) and contact angle (). The following tables show the average values with the errors representing the standard deviation.

S2. Vapor-solid growth of InSb
When FSb is increased to 0.9 Torr, a transition from the VLS growth to the catalyst-free vapor-solid (VS) mode (without any In droplet) occurs, and no more axial growth of InSb is observed. Instead, InSb starts forming a shell around the InAs stem, as we can see in the SEM image S2.1.

S3. High-resolution transmission electron microscopy analysis
We performed TEM analysis of the InAs/InSb NWs grown under different precursor line pressure for the same InSb growth time of 60 min. Figure

S4. X-ray energy dispersive spectroscopy data on the catalyst droplet
The droplet composition for NWs taken from three different samples (grown under different In and Sb line pressure) were measured by energy dispersive X-ray spectroscopy (EDX), with the results summarized in the

S5. ZB crystal phase of InSb segments
Here, we briefly discuss the crystal phase purity of InSb NW sections in the ZB structure from the surface energy point of view (see Figure 5 of the main text for the model parameters). We note that the difference between planar solid-vapor and solid-liquid facets for InSb

  
, as for planar facets, the WZ phase seems to be always enabled around   90 o , as in our case. In fact, this observation was the original argument of Ref. [1], for the prevalence of the WZ phase in VLS III-V NWs. However, these considerations apply to vertical corner facets wetted by the droplet. It was then noticed that III-V NWs may grow with a truncated corner facets which makes nucleation of two-dimensional (2D) islands at the TPL improbable and hence the crystal phase of such truncated NWs should be ZB [2,3]. Therefore, being the ZB phase very predominant in VLS InSb NWs in almost all cases, we speculate that InSb islands always nucleate with truncated lateral facets, as suggested in Ref. [4]. A more detailed analysis of the growth interface of InSb NWs grown by self-catalyzed CBE will be presented elsewhere.

S6. Cooling down experiment
In order to study the effect of the cooling process on the NW morphology and the droplet shape, we have grown two samples using the same parameters (FIn/FSb = 0.2/0.35 and 60 min growth time) but with different growth terminations. In the first case, the sample was cooled down to 150 °C in 3 min under TDMASb line pressure (as for all the other samples reported in the main text, linearly decreasing the line pressure from 0.35 Torr to 0 Torr), while in the second case it was cooled down without any precursor flux. Scanning electron microscopy (SEM) micrographs of the NWs obtained are shown in Figure S6.1. We measured the length and diameter of InSb segments and the contact angle of the droplets of ~ 30 NWs from each sample, following the procedure described in the main text. We found no difference between the two samples in terms of the NW length, diameter and contact angle of the droplet. We obtained L = 180 ± 10 nm, D = 162 ± 3 nm, and = 104°± 2° for the sample cooled down under TDMASb line pressure; and L = 190 ± 10 nm, D = 157 ± 3 nm, and  = 102° ± 2° for the sample cooled down without any flux. Therefore, we concluded that the cooling down process does not affect the morphology of the InSb segment and the In droplet shape. This is reasonable considering that the axial growth will immediately decrease and probably stop during the cooling down step due to the lower temperature and the lower amount of Sb atoms available in vapor phase [5].

S7. InSb length at short growth times
Here we discuss one of the possible explanations of the super-linear axial InSb growth rate for short growth times (< 30 min) observed in the series of samples grown using FIn = 0.2 Torr and FSb = 0.35 Torr (see Figure 1 (a) and (c) of the main text). As known from previously reported TEM analysis of the catalyst-free InAs NWs [6], the NW tip is not perfectly flat, showing instead some inclined facets, resulting in an tapered tip terminating with the flat (111) top facet. The tapered InAs tip shape is also visible in the EDX map of our InAs/InSb NWs (see figure 4 (a) of the main text). However, when we perform SEM imaging if the InAs/InSb NWs, the tapered InAs tip is not visible anymore, suggesting that InSb growth occurs also on the inclined facets of the InAs stem, burying the tapered InAs tip. Indeed, when we measure the InSb segment length (L), we take the InAs/InSb interface as the point at which we see the increase of the NW diameter. Therefore, L can differ from the actual InSb axial segment length (L*), and this will result in an overestimation of the segment length, which is more relevant for short growth times. Figure S7.1 schematically explains this effect: the measured parameter L is given by the actual axial segment length L* plus the height of the tapered InAs tip (h). Since the latter is around 30-40 nm (as measured from the EDX maps), the overestimation of L is more relevant for short growth times (t ≤ 30 min).