Supercycled ALD with In‑Situ O₂ Plasma Tunes ZnO Film Resistivity and Carrier Concentration Across Five Orders

Background

Zinc oxide (ZnO) has long been hailed as a “future material” owing to its superior optical transparency and n‑type conductivity.1–3 Its versatility spans transparent conductive electrodes for flat‑panel displays, touchscreens, low‑emissivity coatings, and thin‑film solar cells,4–6 as well as key components in LEDs, photodetectors, and power devices.7–10 Applications such as transparent resistive random‑access memory (TRRAM) demand ZnO layers with distinct electrical profiles—high‑resistivity films for the switching element and low‑resistivity films for the transparent electrodes.8, 9, 10 Conventional doping, while effective, introduces compositional complexity and potential secondary phase formation.11–14 Consequently, a deposition technique that can modulate the intrinsic electrical properties of undoped ZnO in a single, self‑limiting process is highly desirable. ALD has emerged as a premier method for fabricating high‑quality ZnO with nanometer‑scale thickness control and large‑area uniformity.15–17 Thermal ALD typically employs diethylzinc (DEZ) and water vapor, whereas plasma‑enhanced ALD uses DEZ and O₂ plasma. The electrical tuning in thermal ALD mainly relies on growth temperature adjustments,18, 19 which limit the attainable carrier concentration range. Plasma‑enhanced ALD offers improved control of low‑carrier‑concentration films,20, 21 but suffers from non‑self‑limiting growth at short plasma times, jeopardizing uniformity. Accurate characterization of ZnO electrical properties remains challenging. Hall effect measurements can misinterpret doping mechanisms, while KPFM provides a non‑destructive probe of surface potential and, by extension, Fermi level shifts.22–24 To date, no comprehensive study has correlated KPFM data with ALD‑grown ZnO films. This work fills that gap by demonstrating how a supercycled ALD process—thermal cycles plus in‑situ O₂ plasma—yields ZnO films whose electrical behavior can be fine‑tuned and directly mapped via KPFM.

Methods

All ZnO thin films were deposited on 1 cm × 1 cm SiO₂/Si substrates at 190 °C using a FlexAL OIPT ALD system with DEZ as the Zn precursor. Each supercycle comprises m thermal ALD sub‑cycles (DEZ + H₂O) followed by a single O₂ plasma pulse (Figure 1a). The plasma step employed 60 sccm O₂, 300 W RF power, and 15 mTorr pressure. By adjusting m and the plasma duration (t₃), we controlled film resistivity and carrier concentration. Film thicknesses were measured by spectroscopic ellipsometry (VASE) and modeled with a Tauc‑Lorentz profile. Electrical properties were extracted from Hall measurements (Nanometrics HL5500PC) at 0.5 T, ensuring linear contact between copper probes and the sample. Grazing‑incidence X‑ray diffraction (Rigaku Smartlab, θ = 1°) assessed crystallinity. X‑ray photoelectron spectroscopy (Thermo Scientific Theta Probe) with Al Kα radiation characterized chemical states; C 1s at 284.6 eV served as the energy reference. KPFM imaging was performed on a Nanonics CV2000 using a Pt‑Ir coated tip (65 kHz resonant frequency). Measurements were taken immediately after vacuum chamber evacuation to minimize surface contamination.

a Illustration of one supercycle in the proposed supercycled ALD process. b Growth rates versus O₂ plasma time for supercycled ALD (m = 1) and conventional plasma‑enhanced ALD; dashed lines guide the eye. c Linear fit of growth rate versus thermal cycle count (m) for fixed plasma times (t₃ = 1 s and 8 s).

Results and Discussion

The supercycled ALD process, illustrated in Figure 1a, enables a self‑limiting growth regime while incorporating an in‑situ O₂ plasma treatment. Figure 1b shows that the plasma‑enhanced ALD growth rate rises from ~1.4 to 1.7 Å/cycle as the plasma time increases from 2 to 4 s, then saturates. In contrast, the supercycled ALD growth rate remains ~1.69 Å/supercycle regardless of plasma duration, indicating that the thermal ALD step dominates the film deposition. Figure 1c demonstrates a linear increase in growth rate with the number of thermal cycles (m) per supercycle, with a slope of ~1.67 Å/cycle, consistent with the thermal ALD rate. X‑ray diffraction (Figure 2a) confirms all films crystallize in the hexagonal wurtzite structure, exhibiting a strong ‑axis orientation (I₀₀₂/I₁₀₁ ≈ 2–5). The slight increase in this ratio with longer plasma times suggests a marginal improvement in preferred orientation, yet overall crystalline quality remains stable. Scherrer analysis yields grain sizes of ~11 nm, unaffected by plasma exposure or thermal cycle count. Similar diffraction trends were observed for films grown with varied m values (Figure 2b). Spectroscopic ellipsometry (Figure S2) reveals that optical constants (n, k) and refractive index remain unchanged across different plasma times and thermal cycles, consistent with the preserved crystallinity and morphology (AFM roughness 0.3–0.8 nm, Figure S3). Electrical characterization (Figure 3) shows a dramatic resistivity increase from ~10⁻³ to 10³ Ω·cm as plasma time extends from 0 to 8 s (m = 1). Correspondingly, carrier concentration falls from ~10²¹ to 10¹⁵ cm⁻³, while electron mobility stays around 3.0 ± 1.0 cm²/V·s. This five‑order‑of‑magnitude tuning surpasses our previous plasma‑enhanced ALD approach. Adjusting the thermal cycle count at fixed plasma time (t₃ = 1 s) further refines the resistivity, offering three additional values between 10⁻³ and 10 Ω·cm. KPFM measurements (Figure 4) provide direct evidence of Fermi level shifts. Mean V_CPD values differ by ~0.32 eV between films grown with 0 and 8 s plasma, a significant fraction of the 3.22 eV bandgap. The same trend holds for varied thermal cycles, indicating that changes in electron‑hole balance across the film influence carrier density. We employed the Maragliano model to relate V_CPD to donor concentration (Eq. 2). Using relative V_CPD differences, the calculated carrier concentration ratios (Eq. 3) align closely with Hall measurements (Figure 5), confirming that Fermi level shifts measured by KPFM accurately track carrier density. X‑ray photoelectron spectroscopy (Figure 6) identifies two O 1s components: A (~530.3 eV) attributed to O²⁻ in the ZnO lattice, and B (~532.2 eV) associated with Zn–OH groups. The B peak intensity, a proxy for hydrogen incorporation, diminishes with longer plasma exposure, directly correlating with increased resistivity and reduced carrier concentration (Figure 8). Zn 2p spectra (Figure 7) show no evidence of Zn interstitials or oxygen vacancies, ruling them out as the primary dopants. Collectively, these results support the hypothesis that hydrogen‑related defects act as shallow donors in ZnO. The in‑situ O₂ plasma step selectively removes these defects, shifting the Fermi level and modulating the electrical properties while preserving crystalline and optical integrity.

XRD patterns of ZnO films grown by the supercycled ALD process: (a) varying O₂ plasma times (m = 1) and (b) varying thermal cycles (t₃ = 1 s).

a Resistivity versus plasma time (solid dots) and thermal cycles (open dots). b Carrier concentration and mobility trends.

a–e V_CPD maps for varying plasma times (t₃ = 0–8 s, m = 1). f Average V_CPD values for plasma time and thermal cycle variations.

Carrier concentration ratios derived from Hall and KPFM data as a function of plasma time.

a–e O 1s XPS spectra and Gaussian fits for varying plasma times (m = 1). f Relative Zn–OH peak intensity versus plasma time and thermal cycles.

Zn 2p XPS spectra for films grown with different plasma times (m = 1).

Resistivity and carrier concentration plotted against the Zn–OH peak proportion.

Conclusions

We demonstrate that a supercycled ALD approach, combining thermal ALD with an in‑situ O₂ plasma step, yields undoped ZnO films whose resistivity and carrier concentration can be tuned over five orders of magnitude by adjusting plasma time and thermal cycle count. Hydrogen‑related defects, quantified via Zn–OH XPS signals, are the primary contributors to n‑type conductivity, and their removal by plasma exposure shifts the Fermi level, as confirmed by KPFM. The excellent agreement between KPFM‑derived carrier ratios and Hall measurements validates the use of KPFM as a rapid, non‑destructive diagnostic for ZnO electrical properties. This reliable, scalable technique enables the fabrication of ZnO layers with bespoke electrical characteristics for next‑generation optoelectronic and transparent electronic devices.

Abbreviations

ALD

Atomic layer deposition

DEZ

Diethylzinc

KPFM

Kelvin probe force microscopy

XPS

X‑ray photoelectron spectroscopy

XRD

X‑ray diffraction