Efficient Nitrogen Incorporation in GaP Nanowires via Au‑Catalyzed VLS Growth Using UDMH

Introduction

III–V NWs are widely investigated for applications ranging from optoelectronics to catalysis. Their small cross‑section allows elastic strain relaxation, enabling high‑quality heteroepitaxy even across large lattice mismatches. Consequently, lattice‑matching constraints are relaxed and the focus shifts to engineering optoelectronic, chemical, and structural properties.

Incorporating dilute nitrogen into conventional III–V compounds—forming so‑called dilute nitrides—has been shown to dramatically lower the band‑gap and, for GaP, convert an indirect gap into a quasi‑direct one when the N fraction exceeds ~0.5 %. Additionally, N incorporation enhances aqueous stability, which is critical for photocatalytic applications such as solar water splitting.

Prior approaches to grow N‑containing GaP NWs relied on sublimation‑recondensation or MBE of core–shell structures. Typically, a N‑free core is grown by self‑catalyzed VLS, followed by a dilute nitride shell via conventional layer growth. While these strategies have revealed advantages such as reduced surface recombination and enhanced light harvesting, they also suffer from strong non‑radiative recombination linked to point defects, which are highly sensitive to growth conditions and precursor chemistry.

Au‑catalyzed VLS growth offers superior control over morphology and doping compared to self‑catalyzed growth. However, previous attempts to incorporate nitrogen during Au‑catalyzed VLS growth have failed, largely because the nitrogen precursor inhibited one‑dimensional growth. Here we show that UDMH enables successful N incorporation while preserving the one‑dimensional morphology.

Methods

GaP(N) NWs were grown on GaP(111)B substrates by Au‑catalyzed VLS using a metalorganic vapor‑phase epitaxy (MOVPE, Aixtron AIX 200) system. Trimethylgallium (TMGa), tertiarybutylphosphine (TBP), and unsymmetrical dimethyl hydrazine (UDMH) served as precursors for Ga, P, and N, respectively. Substrates were cleaned with acetone and isopropyl alcohol before deposition of monodisperse Au particles from a colloidal solution. A 15‑min anneal at 550 °C under TBP overpressure formed liquid Au‑Ga droplets. Growth conditions were: TMGa molar fraction χTMGa = 6.16 × 10⁻⁵, TBP/TMGa = 10, temperature 500–550 °C, UDMH/TBP ratio 0–9, duration 16 min, Au particle diameter 50 nm, reactor pressure 50 mbar, total gas flow 3.4 l min⁻¹ (H₂ carrier). Temperatures were measured by a thermocouple inside the graphite suszeptor.

Characterization employed high‑resolution scanning electron microscopy (SEM, Hitachi S 4800‑II). Two samples were further examined by TEM: specimens were mechanically dry‑transferred onto lacey carbon grids and examined with a ThermoScientific Titan³ Themis at 200 kV. Energy‑loss spectra were recorded using the GIF Quantum ERS in diffraction mode (collection angle ~3 mrad) to detect the N‑K edge at 403 eV. Raman spectra were acquired on Si‑substrate‑mounted NWs using a 532‑nm laser (400 µW) focused with a × 50 objective; the signal was detected by a cooled Si CCD.

Results and Discussion

Morphology

Figure 1 displays the morphologies of GaP(N) NWs grown under varying UDMH/TBP ratios and temperatures. No bending or touching was observed immediately after growth; the apparent deformation in the SEM images is due to electrostatic attraction during imaging.

All samples exhibit freestanding, straight, and vertical NWs with homogeneous lengths. In contrast to self‑catalyzed dilute nitride NWs, no parasitic island growth occurs. Temperature and UDMH concentration strongly influence morphology: higher temperature shortens NWs, enhances side‑facet vapor‑solid (VS) growth, and increases tapering. Conversely, increasing UDMH raises the axial growth rate while suppressing the radial rate, thereby reducing tapering. At UDMH/TBP = 9:1, growth becomes unstable, leading to random growth directions and significant length dispersion; surface roughening is also observed, particularly at high temperature and high UDMH supply, likely due to strain from inhomogeneous N incorporation.

Figure 2 summarizes the geometric characteristics as a function of temperature and UDMH/TBP ratio. The axial growth rate (a) rises with UDMH and falls with temperature; the coaxial rate (b) shows the opposite trend. The tapering parameter (c)—defined as (radius_top – radius_bottom)/length—is low for high UDMH ratios and low temperatures, matching the ratio of coaxial to axial growth rates. The total volume (d) is largely unaffected by UDMH concentration, indicating that UDMH does not etch the NWs.

The reduction in tapering with UDMH can be attributed to steric hindrance by partially decomposed UDMH molecules on the side facets, which blocks VS growth while allowing Ga species to reach the Au catalyst and promote VLS growth. Evidence includes the incomplete decomposition of UDMH (5–30 % between 500–550 °C), the high gas‑phase concentration (10–90 × TMGa), and in‑situ spectroscopy showing adsorption of UDMH fragments on the NW surface.

Raman Spectroscopy

Raman measurements were performed on individual NWs with a diameter of ~100 nm, grown for 8 min to minimize parasitic VS overgrowth. For comparison, a lattice‑matched GaPN layer (x = 2.1 %) on Si(100) was also measured. All spectra were normalized to the GaP LO mode.

All spectra display GaP‑like TO (365 cm⁻¹) and LO (399–403 cm⁻¹) phonons. At low UDMH/TBP ratios (0.1–0.3) a surface‑optical (SO) phonon at 397 cm⁻¹ appears, indicative of surface roughness or defects. With increasing UDMH, the SO mode diminishes and a mode near 387 cm⁻¹ (X) emerges, attributed to zone‑boundary LO phonons activated by translational‑symmetry breaking from N incorporation. The N‑related local vibrational mode (NLVM) at ~500 cm⁻¹ scales linearly with substitutional N (x ≤ 2.1 %) and allows estimation of N content.

Deconvolution of the NLVM/LO ratio for the reference GaPN (0.44 ± 0.03) and the 3:1 UDMH sample (0.145 ± 0.028) yields a substitutional N concentration of 0.7 ± 0.2 % for the 3:1 NWs. Lower UDMH ratios produce too weak NLVM signals for reliable quantification.

Interestingly, second‑order Raman scattering (SORS) intensity increases with UDMH, contrary to planar GaPN where it is quenched by lattice distortion. This suggests that N incorporation in NWs reduces stacking‑fault density, thereby lessening lattice distortion and enhancing SORS.

TEM and EELS

TEM and electron energy‑loss spectroscopy (EELS) were employed to corroborate Raman findings and examine crystalline quality. Sample 1A (no UDMH) and sample 1C (UDMH/TBP = 3) were grown at 500 °C.

Both samples exhibit a zinc‑blende lattice with high stacking‑fault densities (150–200 µm⁻¹). However, in sample 1C, 150–300 nm long SF‑free sections are frequently observed, bounded by individual dislocations. The dislocations are mixed‑type, with both screw and edge character, and are likely formed during growth due to the high local strain induced by N incorporation. Their presence appears to suppress subsequent SF nucleation.

Conclusion

We have demonstrated that unsymmetrical dimethyl hydrazine enables efficient nitrogen incorporation into GaP nanowires grown by Au‑catalyzed VLS. Raman spectroscopy confirms increasing N content with higher UDMH supply and shows substitutional incorporation at group‑V sites. Morphologically, UDMH reduces tapering and enhances axial growth, likely through steric hindrance of incompletely decomposed UDMH on the side facets. TEM reveals a zinc‑blende lattice in both N‑free and N‑containing NWs, with a high SF density; yet N‑rich NWs contain extended SF‑free regions bounded by dislocations that suppress SF nucleation. This work establishes UDMH as a viable nitrogen precursor for VLS‑grown nanowires and opens pathways for tailoring their optoelectronic properties.