In situ passivation of Ga x In(1−x)P nanowires using radial Al y In(1−y)P shells grown by MOVPE

Ga x In(1−x)P nanowires with suitable bandgap (1.35–2.26 eV) ranging from the visible to near-infrared wavelength have great potential in optoelectronic applications. Due to the large surface-to-volume ratio of nanowires, the surface states become a pronounced factor affecting device performance. In this work, we performed a systematic study of Ga x In(1−x)P nanowires’ surface passivation, utilizing Al y In(1−y)P shells grown in situ by using a metal-organic vapor phase epitaxy system. Time-resolved photoinduced luminescence and time-resolved THz spectroscopy measurements were performed to study the nanowires’ carrier recombination processes. Compared to the bare Ga0.41In0.59P nanowires without shells, the hole and electron lifetime of the nanowires with the Al0.36In0.64P shells are found to be larger by 40 and 1.1 times, respectively, demonstrating effective surface passivation of trap states. When shells with higher Al composition were grown, both lifetimes of free holes and electrons decreased prominently. We attribute the acceleration of PL decay to an increase in the trap states’ density due to the formation of defects, including the polycrystalline and oxidized amorphous areas in these samples. Furthermore, in a separate set of samples, we varied the shell thickness. We observed that a certain shell thickness of approximately ∼20 nm is needed for efficient passivation of Ga0.31In0.69P nanowires. The photoconductivity of the sample with a shell thickness of 23 nm decays 10 times slower compared with that of the bare core nanowires. We concluded that both the hole and electron trapping and the overall charge recombination in Ga x In(1−x)P nanowires can be substantially passivated through growing an Al y In(1−y)P shell with appropriate Al composition and thickness. Therefore, we have developed an effective in situ surface passivation of Ga x In(1−x)P nanowires by use of Al y In(1−y)P shells, paving the way to high-performance Ga x In(1−x)P nanowires optoelectronic devices.


Introduction
Semiconductor nanowires are promising building blocks for next-generation electronic and optoelectronic nano-devices [1][2][3]. The majority of nanowire devices have been based on unary or binary semiconductors such as Si, Ge, InP, GaAs, InAs, or GaN [4][5][6][7]. In optoelectronic applications, it is necessary to cover a specific range of wavelengths, typically requiring more complex material systems with control over crystal structure, doping profile, and heterostructure formation [8,9]. With the advancement of nanowire research, material systems such as ternary III−V semiconductors and perovskite have been under intensive investigation [10][11][12][13][14][15][16]. Ga x In (1−x) P nanowires, which cover the visible (green) to near-infrared wavelength range, are highly appealing for optoelectronic applications, primarily single and multi-junction solar cells [17,18].
In recent years, the extensive investigation of Ga x In (1−x) P nanowires has led to the demonstration of high-performance nanowire devices [19][20][21][22][23][24]. Due to the large surface-to-volume ratio of nanowires, surface states play an essential role in their optoelectronic properties. The surface states can trap mobile carriers and quench the photoluminescence (PL), resulting in low PL quantum efficiency [25,26]. Nanowires usually have a much shorter PL carrier lifetime as compared to their thinfilm counterparts. Surface passivation is a universal challenge for semiconductor nanowire devices [27][28][29][30]. Passivation schemes for various nanowires have been established, e.g. surface passivation by intentional growth of a co-axial high bandgap shell around the nanowire core [31][32][33][34], wet chemical treatment of nanowire surfaces [35,36], and ex situ deposition of a dielectric layer to the nanowire surface [30,37,38]. However, effective surface passivation methods for Ga x In (1−x) P nanowires are still lacking.
In this study, we studied the surface passivation of Ga x In (1−x) P nanowires by in situ growth of Al y In (1−y) P shell with various bandgap and thickness. Time-resolved photoinduced luminescence (TRPL) and time-resolved THz transmission spectroscopy (TRTS) measurements were performed to characterize the effect of surface passivation. We observed a notable change of the TRPL lifetime and TRTS kinetics in Ga x In (1−x) P nanowires by tuning the Al y In (1−y) P shells' composition and thickness. The hole and electron lifetime of the nanowires with the Al composition of 0.36 in Al y In (1−y) P shells are found to be 40 times and 1.1 times longer than that of the bare nanowires without shells, demonstrating effective surface passivation of hole traps by Al y In (1−y) P shells. The passivation effect decreases when Al composition exceeds 0.36. We correlate this effect with an increase of the coreshell lattice mismatch and a higher level of oxidation and carbon incorporation due to the Al content in the shells. Highresolution transmission electron microscopy (HRTEM) images of these samples revealed defects between the nanowire core and the high Al-content shells. Furthermore, by varying the thickness of Al y In (1−y) P shells, we concluded that both the hole and electron trapping and the overall charge recombination in Ga x In (1−x) P nanowires can be substantially passivated by growing an Al y In (1−y) P shell with an appropriate Al composition and thickness. Therefore, our study demonstrates an effective in situ surface passivation method for Ga x In (1−x) P nanowires using radially-grown Al y In (1−y) P shells.

Methods
Ga x In (1−x) P nanowire arrays were grown by the use of metalorganic vapor phase epitaxy (MOVPE) from Au seed particles patterned on InP (111)B substrates. Two sets of samples were produced. We first grew and characterized a set of Ga 0.41 In 0.59 P/Al y In (1−y) P core-shell samples for which the Al composition of the shell was varied. The total NW length was grown to approximately 2 μm which is monitored in situ by an EpiR DA UV optical reflectance setup from LayTec AG. Further details of the reflectance spectrometry can be found in the electronic supplementary material (ESM) (available online at stacks.iop.org/NANO/32/425705/mmedia). The Al y In (1−y) P shell composition was controlled by tuning the TMAl/TMIn molar fraction ratio. The shell growth time was kept constant. The sample details, together with optical measurement results, are summarized in table 1. The second Table 1. The first set of samples: Ga 0.41 In 0.59 P nanowires without/with Al y In (1−y) P shells of different Al composition. The Ga composition of the Ga 0.41 In 0.59 P core was determined by room-temperature PL. The Al composition of Al y In (1−y) P shell was obtained from EDX measurement. According to the literature, the energy band offset between the Ga 0.41 In 0.59 P core and Al y In (1−y) P shell was calculated [39][40][41]. The average shell thickness was measured according to HRTEM images. set of Ga 0.31 In 0.69 P/Al 0.22 In 0.78 P core-shell samples, which are grown with constant shell composition but different shell growth time, were intended for examining the effect of shell thickness on surface passivation. The shell thickness in the second set of samples was varied from 10 to 35 nm, as shown in table 2. Growth details can be found in S1(1) and table S-1 in the ESM. The morphology of nanowires was inspected by using an LEO 1560 field-emission scanning electron microscopy (SEM). X-ray diffraction (XRD) measurement provides information on material composition. The samples' crystal structure and the material composition were analyzed by using a JEOL 3000F HRTEM equipped with energy-dispersive x-ray spectroscopy (EDX). TRPL and TRTS measurements were performed to reveal the charge recombination processes of the nanowires, with details given in S1(2) and S1(3) in the ESM. and (e), nanowire bending and kinking occurs when the shell's Al composition is higher than 0.48. The bending of nanowires is a clear indication of considerable stress over nanowires caused by the relatively high core-shell lattice mismatch and the facet selective nucleation and growth [42,43]. Distinct XRD peaks for the InP substrates and Ga x In (1−x) P nanowires indicate a 35%±10% Ga composition (figure S-1(a) in the ESM) [44]. The XRD peaks of different samples are overlapped largely, indicating a reproducible epitaxial growth.

Results and discussions
To assess the optical properties of the samples, TRPL measurement was performed at room temperature. As shown in figure S-1(b) in the ESM, the time-integrated TRPL spectrum shows the emission peak of GaInP at ∼1.74 eV, corresponding to a Ga composition x of ∼0.41. The PL peaks are relatively broad related to an inhomogeneous composition distribution within an individual nanowire and different nanowires. We made an effort to measure the PL of Al y In (1−y) P shells by increasing the excitation intensity. However, no PL emission signal of the shells was observed within the detection capability of the PL setup. Several factors can cause the negligible PL emission of Al y In (1−y) P shells. Firstly, the bandgap of Al y In (1−y) P turns indirect when the Al composition x approaches or exceeds ∼0.44 preventing efficient radiative recombination [45]. Secondly, photogenerated free carriers in the Al y In (1−y) P shells can be trapped by the shell's surface states, thus quenching the PL. Thirdly, the charges in the shell could tunnel to the nanowire core. In order to obtain the shell composition, we performed EDX measurement in TEM equipment. We observed variations of the Al composition and shell thickness, both radially and axially, over an individual nanowire. The composition variation is originated from different gas and surface diffusion length among Al atoms, In atoms, and their precursors, together with the fact of phase segregation of Al y In (1−y) P, as reported in previous studies [42,46]. The average Al composition of the shells of each sample can be found in table 1.
In order to investigate the effect of Al y In (1−y) P shells on carrier recombination dynamics of the Ga 0.41 In 0.59 P nanowires, TRPL kinetics of the nanowires were measured, as shown in figure 2. The main carrier recombination processes  in semiconductor nanowires include Auger recombination, bimolecular recombination, and charge trapping. Typically, a significant contribution of Auger recombination or bimolecular recombination processes should lead to a faster PL decay with increasing excitation fluence, as they are the third and second-order recombination processes, respectively. However, we observed that the PL of the core nanowires decayed slower under increasing excitation fluence (figure S-2 in the ESM), indicating that PL kinetics are probably dominated by charge trapping processes in the excitation fluence range of the measurements [47][48][49]. We noted that TRPL decays in figure 3 were non-monoexponential, which could be related to an inhomogeneous distribution of traps in the nanowires [50][51][52]. The TRPL lifetime was quantified using the 1/e methods, and the results are shown in table 1. We found that the TRPL lifetime increased with Al composition, reaching a peak when Al composition is 0. 36 photoconductivity, which is determined by the product of mobility and concentration of mobile photogenerated charges [47]. In bulk Ga x In (1−x) P, the electron mobility is significantly higher than holes. Thus, we presume the photoconductivity mainly arises from photogenerated electrons, therefore selectively assess the population of mobile electrons. Figure 3 shows the photoconductivity kinetics of Ga 0.41 In 0.59 P core with various Al y In (1−y) P shells. We observed that the photoconductivity (Δσ) decayed nonexponentially, supporting the previous conclusion of the inhomogeneity of the nanowires.
To assign the Δσ decay processes, various excitation fluences were used for TRTS measurements (figure S-3 in the ESM). The Δσ decay slows down (or does no changes) with the increase of the excitation fluence. Similar to the considerations on the PL decays, TRTS decay is due to the trapping of electrons together with trap filling. Assuming that the electron mobility does not change much in the measured time range, the decay of TRTS kinetics of Ga 0.41 In 0.59 P nanowires can be attributed to charge recombination and the electron trapping processes.
The TRTS kinetics are fitted by double exponential decay functions (table S-3 in the ESM). The averaged TRTS lifetime of bare Ga 0.41 In 0.59 P nanowires, Ga 0.41 In 0.59 P/Al 0.36 In 0.64 P core-shell nanowires, and Ga 0.41 In 0.59 P/Al 0.55 In 0.45 P coreshell nanowires are 1610 ps, 1810 ps, and 470 ps, respectively. The TRTS lifetime of these nanowire samples is much longer than their TRPL lifetime. This suggests that the hole trapping dominates the PL decay. Comparing the TRPL and TRTS lifetime in bare Ga 0.41 In 0.59 P nanowires and Ga 0.41 In 0.59 P/Al 0.36 In 0.64 P core-shell nanowires, the hole and electron lifetime of Ga 0.41 In 0.59 P/Al 0.36 In 0.64 P core-shell nanowires are approximately 40 times and 1.1 times larger than that for the bare core nanowires, respectively. This suggests that the main charge trapping process in  Ga 0.41 In 0.59 P nanowires that can be passivated by Al y In (1−y) P shell is the hole trapping. When the Al content of the shells is increased, both PL and photoconductivity decay become shorter. By comparing TRPL and TRTS lifetime of Ga 0.41 In 0.59 P/Al 0.36 In 0.64 P and Ga 0.41 In 0.59 P/Al 0.55 In 0.45 P core-shell nanowires, we found that the hole and electron lifetime reduced by about 35 and 3.8 times, respectively. This indicates that excessive Al content of shells creates new electron and hole traps. Note that the distorted nanowire top of the high-Al-content samples can also be a cause of the deterioration of TRTS kinetics [48].
To identify the nature of the traps, we performed HRTEM measurements on selected samples with low or high Al-content shells to inspect their microstructure, as shown in figure 4. The HRTEM images show that both low and high Al-content shells follow the same crystal structure as that of the nanowire core, i.e. zinc-blende structure with twin planes. As shown in figures 4(c) and (d), inclined defects (along the diagonal (111)B direction) occur through the nanowire bottom to top for the high Al-content shell. Some polycrystalline or amorphous areas can be observed at the zigzag structures' concave positions at the shell surface due to oxidation. These phenomena are probably related to the strain between core and shell because of lattice mismatch. Further investigation is needed to identify its origin. In any case, these defects can (and most likely do) behave as trap centers for carriers that quench the PL and reduce the TRTS lifetime.
For the first set of samples, we noted that both the Al composition and shell thickness are different between Samples 4 and 5. To identify the effect of shell thickness and further optimize the passivation, the second set of samples with various shell thickness were grown. Room-temperature PL measurement shows an emission peak at ∼1.61 eV, corresponding to a Ga composition of 0.31. EDX measurement reveals an average shell content of Al 0.22 In 0.78 P.
As shown in figure 5, we observed an increase of the TRPL lifetime with the increase of the shell thickness. The PL decay slows down drastically with an increase of the shell thickness up to 17 nm, whereas the lifetime increase is less dramatic with the further increase of the thickness. We noted that the photoconductivity of the sample with a shell thickness of 23 nm decays 10 times slower compared to that of the bare core nanowires (figure S-4 in the ESM). Meanwhile, the slow decay of all samples looks similar, which may contain further information worth investigation in the future. Combining the results of TRPL and TRTS, we can conclude that both the hole and electron trapping and the overall charge recombination in Ga x In (1−x) P nanowires can be substantially passivated by growing Al y In (1−y) P shells with an appropriate Al composition and thickness.

Conclusions
We have passivated Ga 0.41 In 0.59 P nanowires by growing Al y In (1−y) P shells in situ using MOVPE. The nanowires' hole and electron lifetime with the Al 0.36 In 0.64 P shell are 40 times and 1.1 times smaller than that of the bare nanowires without shells, demonstrating effective surface passivation by Al y In (1−y) P shells. For the nanowires covered by the shells with higher Al composition, both PL and photoconductivity decay rates increase with the Al composition of the shell, which can be attributed to an increase of the traps and defect formation. Moreover, we observed that a certain shell thickness of approximately ∼20 nm is needed for efficient passivation of Ga 0.31 In 0.69 P nanowires. The photoconductivity of the sample with a shell thickness of 23 nm decays 10 times slower compared with that of the bare core nanowires. In summary, our results show an effective in situ passivation to Ga x In (1−x) P nanowires, achieving a pronounced improvement of the optical properties, and pave the way to the application of Ga x In (1−x) P nanowires in high-performance optoelectronic devices.