Measuring Local Open‑Circuit Voltage in Si Nanowire Radial Junctions Using Kelvin Probe Force Microscopy

Introduction

Semiconductor nanostructures, especially nanowire arrays with radial junctions, promise significant gains in light trapping and carrier collection, potentially surpassing planar architectures [1, 2]. Silicon nanowire RJs have already achieved efficiencies approaching 9.6 % when fabricated on thin‑film platforms [3]. However, evaluating the photo‑electrical behavior of individual nanowires remains a critical challenge for device optimisation.

We employ KPFM to probe local V_OC on RJ SiNWs. While KPFM‑derived V_OC has been validated on planar photovoltaics [3, 4], its application to nanostructured devices is non‑trivial because it requires separate dark and illuminated measurements to capture the surface photovoltage (SPV).

Our first strategy examines completed RJ SiNW cells that include an ITO front contact. We compare conventional current‑voltage (I‑V) data with KPFM‑derived SPV across a range of illumination powers. The second strategy focuses on isolated RJ SiNWs (no ITO) to study how tip geometry and illumination angle affect SPV extraction. Finally, macroscopic Kelvin probe measurements provide spectral SPV data (SPS) on both isolated and completed devices.

Materials and Methods

SiNW Growth and Radial P‑I‑N Junction Device Fabrication

RJ SiNWs were grown on ZnO:Al‑coated Corning glass (Cg) using plasma‑enhanced chemical vapor deposition (PECVD) at 500 °C with Sn nanoparticle catalysts. A conformal intrinsic layer (80 nm) followed by an n‑type layer (10 nm) of hydrogenated amorphous silicon (a‑Si:H) was deposited at 175 °C onto the p‑type SiNW core. Completed devices were finished with a 4 mm‑diameter ITO top contact deposited by sputtering under a mask. Full fabrication details appear elsewhere [1, 5–7].

Kelvin‑Probe and Surface Photovoltage

We use amplitude‑modulation (AM) KPFM for its superior stability over frequency‑modulation (FM) when scanning topography‑rich nanowires. Measurements were performed on a HORIBA/AIST‑NT TRIOS platform featuring a 1310 nm laser‑beam‑based deflection system (LBBDS) to minimise photo‑excitation artifacts [8–10]. The system offers three objectives for top, side, and bottom illumination.

SPV is obtained by subtracting the dark CPD from the illuminated CPD, a technique proven for V_OC extraction [5, 11]. Illumination is provided by an OXXIUS 488 nm laser diode with variable power.

Two conductive AFM tips were used: ARROW‑EFM (standard tip) and ATEC‑EFM (tilted tip). Both have doped silicon cantilevers coated with PtIr.

Macroscopic Kelvin probe measurements with a 2 mm steel tip and a halogen‑lamp monochromator (400–1000 nm) complement the nanoscale data, enabling qualitative surface‑photovoltage spectroscopy (SPS).

Macroscopic I‑V Measurements Combined to KPFM

For completed devices, a KEITHLEY 2450 SourceMeter and a tungsten needle positioned via a micro‑positioner were used for I‑V characterization under the same AFM setup. Both dark and illuminated (488 nm, 70–1000 µW) I‑V traces were recorded.

Schematics of the measurement setup for both KPFM and macroscopic I‑V measurements

KPFM mapping on isolated RJ SiNWs was performed with both ARROW and ATEC tips under top and side illumination at identical power levels.

Results and Discussion

We first evaluated the influence of the AFM LBBDS wavelength. Macroscopic I‑V curves taken with the TRIOS (1310 nm) and an external AFM (690 nm) show negligible differences in the dark, with a 0.5 mV shift in V_OC and 1 nA in I_SC when the 1310 nm beam is active. In contrast, the 690 nm beam induces a pronounced photovoltaic response (545 mV V_OC, 28 µA I_SC), underscoring the importance of a near‑infrared LBBDS for dark measurements.

I‑V curves of a SiNW RJ device under dark (black), 1310 nm LBBDS (blue), and 690 nm LBBDS (red). The inset zooms the linear region (−5 mV to +5 mV).

Figure 3 illustrates I‑V characteristics under 70, 150, 270, and 560 µW illumination, showing the expected increase of I_SC and V_OC with light intensity. The accompanying KPFM map displays topography, dark CPD, and illuminated CPD (488 nm, 270 µW). CPD variations (~±10 mV) at nanowire edges are attributed to topography‑induced artifacts; values extracted from the nanowire tops are used for subsequent analysis.

a Macroscopic I‑V curves at 66, 5, 149, 268, and 555 µW (488 nm); b topography, dark CPD, and illuminated CPD (270 µW).

Figure 4 compares V_OC from I‑V and SPV from KPFM versus illumination power for two devices. The maximum deviation is <5 % at 70 µW and <2 % at higher powers. Both curves follow a logarithmic trend (500–600 mV) with ideality factors of 1.5 ± 0.1 (device 1) and 1.75 ± 0.25 (device 2), consistent with literature for a‑Si:H P‑I‑N junctions [12–14].

V_OC and SPV versus light power for devices 1 (a) and 2 (b).

SPV measurements on isolated RJ SiNWs highlight shadowing effects. Using ARROW tips with top illumination yields low SPV values (Fig 5a, squares) and a slope ~0.4. Switching to a tilted ATEC tip increases SPV by 40 % (Fig 5b, triangles). Switching illumination from top to side with an ARROW tip (Fig 5c, blue dots) further improves SPV, yet values remain below those of completed devices. Shadowing is absent in completed cells because the ITO layer homogenises the photovoltage over many nanowires.

SPV versus light power on isolated RJ NWs with ARROW‑EFM and ATEC‑EFM tips under top and side illumination. Reference device 1 is shown. Bottom right: topography of isolated NWs.

Qualitative SPS on bundled isolated devices (Fig 6a) reveals significant SPV (80–260 mV) in the near‑infrared (800–1000 nm), contrasting with the negligible SPV of completed devices below 800 nm. In contrast, a completed device shows a sharp SPV rise at 630 nm, reaching 560 mV, but drops below 10 mV in the NIR. Removing the ITO layer from a completed device drastically reduces SPV (400–750 nm) immediately after etching, but the signal recovers after 72 h, approaching the level of isolated NWs. These findings confirm that the ITO/a‑Si:H interface is crucial for high V_OC; its absence limits band‑bending and SPV magnitude.

SPS measurements: a completed device vs. isolated NWs; b completed device before and 72 h after ITO removal.

Overall, local SPV mapping with KPFM provides a reliable proxy for V_OC in fully fabricated RJ SiNW devices, whereas isolated NWs exhibit reduced SPV due to tip shadowing and the lack of an ITO contact.

Conclusion

Simultaneous I‑V and KPFM studies on completed RJ SiNW cells confirm that SPV extracted via KPFM closely matches V_OC from conventional measurements. In contrast, isolated RJ SiNWs show reduced SPV due to tip shadowing; optimizing tip shape and illumination direction mitigates this effect but does not fully recover the V_OC of completed devices. SPS experiments demonstrate that the ITO/a‑Si:H interface is essential for high photovoltage; its absence in isolated NWs limits band‑bending and SPV. Nevertheless, the linear correlation between local SPV and device V_OC validates contactless, nanoscale SPV as a powerful tool for assessing photovoltaic performance of individual SiNW junctions.

Availability of Data and Materials

The datasets used and/or analysed during this study are available from the corresponding author upon reasonable request.

Abbreviations

AFM

Atomic force microscopy

AM

Amplitude modulation

a-Si:H

Hydrogenated amorphous silicon

Cg

Corning glass

CPD

Contact potential difference

FM

Frequency modulation

ITO

Indium‑tin‑oxide

I-V

Current‑voltage

KPFM

Kelvin probe force microscopy

LBBDS

Laser beam‑based deflection system

n

Ideality factor

NW

Nanowire

PECVD

Plasma‑enhanced chemical vapor deposition

PV

Photovoltaic

RJ

Radial junction

SiNW

Silicon nanowire

SPS

Surface photovoltage spectroscopy

SPV

Surface photovoltage

V_OC

Open circuit voltage