Wurtzite AlGaAs Nanowires

Semiconducting nanowires, unlike bulk, can be grown in both wurtzite and zincblende crystal phases. This unique feature allows for growth and investigation of technologically important and previously unexplored materials, such as wurtzite AlGaAs. Here we grow a series of wurtzite AlGaAs nanowires with Al content varying from 0.1 to 0.6, on silicon substrates and through a comparative structural and optical analysis we experimentally derive, for the first time, the formula for the bandgap of wurtzite AlGaAs. Moreover, bright emission and short lifetime of our nanowires suggest that wurtzite AlGaAs is a direct bandgap material.


Results and Discussion
In Fig. 1a,b, we show the transmission electron microscopy (TEM) images of a typical Al 0.6 Ga 0. 4 As nanowire. In Fig. 1a, only a few short zincblende insertions, spontaneously formed during the nanowire growth, are visible as darker lines, consistent with our previous observation of a typical density of ~10-30 insertions/µm 13 . The zoom-in in Fig. 1b is optimized, during the TEM measurement, to resolve the inner part of the nanowire, highlighting the core-shell structure. The thick shell surrounding the core is masking the difference in the Al content between the core and the shell (Fig. 1a,b), resulting in a similar core-shell brightness in the images. Furthermore, the shell, protects and passivate 18 the core surface. Given a certain Al/Ga ratio in the reactor, the nanowires grow with such core-shell formation spontaneously, where the core has lower Al content than the shell, and thus is expected to have lower bandgap. The origin of such growth has been studied in our previous work 17 , where the core-shell composition has been measured, via energy dispersive X-ray spectroscopy (EDX), to be always lower than nominal. The core grows with the diameter equal to the size of the catalyst droplet initially deposited on the substrate, which is kept constant during growth. Furthermore, TEM diffraction pattern analysis shows the nanowires to be always mostly pure wurtzite structure. The core-shell structure results in a double-peak emission, shown in the photoluminescence (PL) measurement in Fig. 1c. Here, the sample is excited above-bandgap and is scanned while continuously collecting micro-PL signal for 30 s, thus integrating the emission of large number of nanowires, on the order of thousands. This results in a measurement similar to macro-PL, referred here as scanning-PL (sPL).
Because of the measured difference in Al content in the core and the shell, we attribute the short-wavelength and long-wavelength branches of the emission to shell and core, respectively. Further details on TEM, EDX and sPL can be found in our previous works 13,17 .
In Fig. 2a,c, we show the sPL measured at different excitation powers for samples with Al content of 0.6 and 0.3, respectively. The long-wavelength peak, attributed to the core, is dominant at low excitation power. With increasing excitation power, the shell states start to populate and emit, resulting in the emergence of a short-wavelength peak that eventually dominantes. We attribute this behavior to state-filling effects in the thin core (~20 nm) at high excitation powers. The excitation power dependent PL measurements (see Fig. S1 in the Supplementary Information) show this dynamics and also high brightness of our nanowires. We note that the data in Fig. 2a,c are normalized to their relative maxima and none of the peaks saturate. In Fig. 2b,d, we show the micro-PL for five different individual nanowires on each of the samples. This shows that the integrated emission measured with the sPL at low excitation powers is composed of a diverse and rich ensemble of individual nanowire peaks. These distributed peaks are attributed to crystal-phase structures in the nanowires with type-II band alignment 6,19 . These formations confine the charges and provide a radiative path for recombination. At the same time, they are consistently observed to be only a few atomic-layers thick, and thus are expected to have transition energies very close to the bandgap of AlGaAs. We note that previous works 20 have highlighted the possibility of having random aggregates of Al-poor regions at the corners of hexagonal GaAs/AlGaAs structures, emitting light with similar spectral features to what observed in Fig. 2b,d. Our cross-sectional TEM measurements (not shown) do not reveal any trace of such compositional fluctuations and thus we attribute the emission to crystal phase structures, which are indeed observed in our TEM data.
By measuring the nanowire ensemble emission (such as shown in Fig. 2a,c) for samples with different Al contents, we obtain the dependence of the nanowire emission energy on the actual Al content and thus experimentally determine the bandgap of wurtzite Al X Ga 1-X As for 0.1 < x < 0.6. In Fig. 3a we show the integrated emission of a large number of nanowires, for different Al contents. All the spectra show two peaks, due to core and shell emission, except for the sample with nominal Al 0.1. In this case the composition of both the core and the shell is very close to the nominal. This results in a non-distinguishable core in TEM observations and in only one peak in the emission spectrum. We note that at low excitation power, crystal-phase structures can prevent the appearance of band-to-band transitions in the emission spectra, as photo-generated carriers fall into the lower energy states in the crystal-phase dots. However, we consistently observe that crystal phase insertions are very small in size -a few monolayers only (see Fig. S2 in the Supplementary Information). Thus the energy levels in such structures are very close to the band-to-band transition energies of the host material. Macro-PL, on the other hand, showing the average distribution of all the emission lines, is clearly grouped in two peaks -core and shell. We thus have used macro-PL (or sPL) data for the derivation of AlGaAs bandgap.
The position of the maxima in the spectra in Fig. 3a is evaluated in respect to the actual content of Al measured via EDX 17,21 , and is shown in Fig. 3b, revealing the linear trend. In this case, the shell and core emission peak energies are plotted against different values of x. The values of measured Al composition and corresponding energy of emission are summarized in Table 1. The agreement of the two peak emission with a linear trend confirms our We also note no evident change of the slope, known for zincblende AlGaAs at Al content of ~0.4 due to the bandgap transition from direct to indirect 23 . The lack of such transition for the wurtzite phase suggests to extend the range of emission wavelength that can be reached with AlGaAs. Furthermore, we note that even if the experimental data follows a linear trend for 0.1 < x < 0.6, the bandgap is not necessarily linear. Thus linear extrapolations of our formula above x = 0.6 might not lead to correct predictions. Comparison with current predictions 24 and experiments 25 indeed suggests the non-linearity of the bandgap. In Fig. 3b, we have also included an estimated uncertainty of 0.1 in Al content by means of a red shade. This uncertainty is the estimated maximum uncertainty in Al content due to statistical fluctuations of the growth parameters, resulting in small variations of Al content from nanowire to nanowire, as we observe in our TEM measurements. In Fig. 4a we show the temperature dependence of the photoluminescence of a Al 0.3 Ga 0.7 As nanowire ensemble. Our nanowires show pronounced emission at room temperature, even though the intensity drops in the range 200-300 K. The wavelength of emission does not have a noticeable change in the temperature range from 2 K to about 100 K, and undergoes a red-shift at higher temperatures. This behavior is expected from the Varshni empirical equation for the temperature dependence of semiconductor bandgaps 26 and is similar, for instance, to the wurtzite GaAs 22 . The measurements at low temperatures (2-50 K) show a notched spectrum, composed of several sharp features, that we attribute to bright emission from crystal-phase structures collected during the scan at these temperatures. These sharp features disappear at higher temperatures, leaving smooth spectra. We www.nature.com/scientificreports www.nature.com/scientificreports/ associate this to the escaping of charges from the trapping crystal-phase structures due to thermal energy, allowing for emission from the AlGaAs bandgap transition.
In Fig. 4b, we show the lifetime measurement for the emission peak at 770 nm at 200 K. The fit to a monoexponential decay reveals a fast decay with lifetime τ = 773 ± 17 ps, similar to the room temperature decay rate of GaAs 27 . These results further suggest that the emission at high temperature originates from the direct bulk-like emission of wurtzite AlGaAs and not from crystal-phase related emission, which is expected to have spatially indirect transitions with slower recombination times 6 . Although the actual lifetime in our measurement can be masked by non-radiative recombinations expected at this temperature, together with high brightness of the nanowires it suggests that wurtzite AlGaAs is a direct bandgap semiconductor for Al content less than 0.6.
Finally, in order to exclude a possible emission from the AlGaAs layer in between nanowires, we transfer the nanowires on a clean silicon-dioxide substrate. In Fig. 4c, we compare the emission spectra from an ensemble of as-grown and transferred nanowires. The clear similarity of the spectra leads to conclude that none of the emission studied in this work is due to the remnant AlGaAs layer.
In summary, we have experimentally derived a formula for the bandgap of wurtzite AlGaAs for aluminum contents up to 0.6. Our results show that the bandgap, within that range, follows a linear trend and agrees remarkably well with known experimental values for Al content of 0, i.e. wurtzite GaAs. We reported a nearly-pure crystalline structure, via TEM observations, and showed its wide tunability range of the emission wavelength via PL measurements. Finally, we have shown the spectra of the emitted light as a function of sample temperature, observing the expected trend for bandgap of bulk semiconductors. High brightness and short lifetime of our nanowires indicate that the bandgap of Wurtzite AlGaAs is direct for Al content less than 0.6, although an unambiguous proof is yet to be demonstrated.

Methods
Sample growth. Our nanowires are grown by Au-catalyzed vapor-liquid-solid (VLS) method in a molecular beam epitaxy reactor on a Si(111) substrate. First, an Au layer 0.1 nm thick is deposited at 550 °C. The gold forms droplets of ~20 nm in diameter which define the size of the core of the nanowires. Al, Ga and As molecular beams are applied simultaneously to grow Al X Ga 1-X As nanowires at 510 °C. The ratio between the Al and Ga defines the nominal Al content x. MBE growth of Au-catalyzed AlGaAs nanowires leads to spontaneous formation of an   As nanowires compared to nanowires transferred to a silicon-dioxide substrate. The similar spectra show that the measured emission comes from the nanowires and not from any layers residual from growth. The spectra are vertically shifted for clarity.