Superior Light Confinement in Hemiellipsoid‑Modified GaAs Nanowire Arrays with Minimal Effective Thickness

Background

Photovoltaic (PV) technology has evolved rapidly, yet the cost of crystalline silicon modules remains a barrier to mass deployment. Thin‑film devices offer lower material consumption but suffer from limited light absorption due to their reduced optical thickness. Light‑management strategies such as antireflection coatings and surface texturing can mitigate this, yet they add complexity and expense. Nanostructured absorbers—nanowires, nanocones, nanopits, nanohemispheres—provide intrinsic optical benefits, including graded refractive index and guided resonances, without requiring additional layers.

Despite these advantages, no single nanostructure optimally balances broadband absorption and fabrication simplicity. We therefore explore a hybrid approach that combines a top hemiellipsoid or inverted hemiellipsoid cap with a GaAs nanowire array. This design leverages the low‑index contrast at the nanowire apex to suppress reflection while the cap geometry scatters light into guided modes, achieving high absorption with a minimal effective thickness.

Methods

Arrays were modeled on a square lattice with a 600 nm period. Each unit cell comprised a GaAs nanowire of diameter D, total height H, and a cap height h (see Fig. 1). The finite‑difference time‑domain (FDTD) method solved Maxwell’s equations with periodic side boundaries and perfectly matched layers on the top and bottom. Reflection (R), transmission (T), and absorption (A = 1 − R − T) were extracted across 300–1000 nm, covering the GaAs bandgap (1.42 eV) and the bulk of the solar spectrum.

a Schematic of a hemiellipsoid‑modified NW array, and b a unit of an inverted hemiellipsoid‑modified NW array for optical simulations. The parameters D, H, and h are labeled.

GaAs was chosen because of its 1.42 eV bandgap and high carrier mobility. To quantify absorption enhancement, we calculated the normalized theoretical photocurrent density ^NJ_ph, defined as the ratio of the structure’s ideal photocurrent to that of a perfect absorber (≈32.0 mA cm⁻² at AM 1.5 G). A value of 1.0 corresponds to 100 % collection of photons above the bandgap.

Results and Discussion

Figure 2 illustrates ^NJ_ph versus cap height h for arrays with H = 1000, 2000, and 3000 nm and diameters 100, 300, and 500 nm. Small diameters (100 nm) show a monotonic decline in ^NJ_ph as h increases, because the low effective refractive index limits light confinement and the cap enhances transmission. For larger diameters (300–500 nm), an optimal h exists that maximizes absorption: for D = 500 nm and H = 1000 nm, inverted hemiellipsoid caps of 1000 nm and 750 nm height achieve ^NJ_ph of 0.90 and 0.95, respectively, with effective thicknesses of only ≈180 nm and ≈270 nm.

Normalized theoretical photocurrent density for hemiellipsoid and inverted hemiellipsoid‑modified GaAs NW arrays as a function of cap height h at total heights 1000 (a), 2000 (b), and 3000 nm (c). Diameters are 100, 300, and 500 nm. Red dotted and dashed lines mark ^NJ_ph of 0.90 and 0.95.

Antireflection is inherent to NW arrays because their effective refractive index gradually matches that of air. However, for thin wires (100 nm) the low fill factor leads to high transmission, especially in the long‑wavelength regime (Fig. 3a). Adding a cap does not improve reflection but increases transmission, reducing overall absorption (Fig. 3b). The dominant mechanism here is the HE₁₁ leaky mode, as illustrated in the inset of Fig. 3b.

a Reflection/transmission and b absorption of arrays with H = 2000 nm and D = 100 nm. c Reflection, d transmission, and e absorption of arrays with H = 2000 nm and D = 500 nm. f Absorption of pure NW arrays with diameters 100, 300, and 500 nm and H = 2000 nm. Insets show electric‑field distributions for the HE₁₁ mode and the 500 nm wire at 810 nm.

For larger diameters (300–500 nm), the fill factor increases, raising the effective refractive index and reflection. Here, the hemiellipsoid or inverted cap dramatically reduces reflection and boosts absorption (Figs. 3c, e). The optimal h range is broad: for D = 500 nm, inverted caps between 350–2000 nm and hemiellipsoid caps between 600–2000 nm maintain ^NJ_ph > 0.95. Excessively tall caps, however, enhance transmission near the bandgap and diminish absorption, producing the characteristic rise‑then‑fall trend seen in Fig. 2.

Figure 3f displays absorption spectra for pure NW arrays with the three diameters. As D increases, the absorption edge shifts to longer wavelengths and the dominant mechanism transitions from leaky modes to scattering. Oscillations around 800 nm for 500 nm wires arise from guided longitudinal resonances, evident in the inset.

Spatial distribution of the photo‑generated carrier generation rate under AM 1.5 G for arrays with H = 2000 nm and D = 500 nm, modified by (left) hemiellipsoids (h = 500 nm) and (middle) inverted hemiellipsoids (h = 500 nm). The generation rate in the pure NW array is shown on the right for comparison.

Top caps expand the volume over which carriers are generated, improving carrier collection—especially for planar p‑n junctions—and increasing tolerance to bulk defects. However, the cap also raises surface carrier density, underscoring the need for effective surface passivation to suppress recombination.

Figure 5 demonstrates that both hemiellipsoid and inverted caps maintain high absorption even at oblique incidence. At 60°, absorption drops by less than 5 %, and the theoretical photocurrent density remains within 5 % of the normal‑incidence value, confirming excellent omnidirectional performance.

Absorption spectra of hemiellipsoid‑ (a) and inverted hemiellipsoid‑modified (b) GaAs NW arrays (H = 2000 nm, D = 500 nm, h = 500 nm) at incident angles 0°, 30°, and 60°. Insets list theoretical photocurrent densities for each angle.

To validate the simulation, we compared circular‑cross‑sectioned wires with realistic orthohexagonal ones (same volume, 2 µm length). Absorption spectra overlapped across 310–873 nm, confirming that the findings apply to experimentally fabricated wires with non‑circular cross‑sections.

Comparison of absorption spectra for GaAs NW arrays with circular and orthohexagonal cross‑sections. Period and wire length are 600 nm and 2 µm, respectively. Wire volumes are matched by the equivalent diameter (100, 300, 500 nm).

These results illustrate that a simple top‑cap modification, coupled with an appropriately sized nanowire base, provides a practical route to high‑efficiency, low‑thickness light absorbers.

Conclusions

We have demonstrated that adding hemiellipsoid or inverted hemiellipsoid caps to GaAs nanowire arrays dramatically improves light confinement while keeping the effective thickness below 300 nm. The enhanced antireflection and scattering synergistically boost absorption, achieving 90–95 % capture of photons above the bandgap with cap‑height‑controlled effective thicknesses of ≈180 nm and ≈270 nm. The modified arrays also exhibit robust omnidirectional absorption and an expanded carrier‑generation profile, which is advantageous for planar p‑n devices. These findings provide a straightforward design principle for advanced, low‑cost solar absorbers.