Unraveling the crucial role of trace oxygen in organic semiconductors

Organic semiconductors (OSCs) are attractive for next-generation optoelectronics due to low cost, mechanical flexibility, and large-area processing. Their weak intermolecular interactions and limited orbital overlap render them highly susceptible to extrinsic defects and impurities. As a result, measured properties often reflect a combination of intrinsic characteristics and extrinsic factors, including those at ultralow (trace) levels that can dominate behavior. Prior work has identified critical roles of extrinsic species such as silanols inhibiting n-type conduction, isomers affecting luminescence, and water-induced charge trapping. Among ubiquitous extrinsic factors, oxygen is the most prevalent impurity and has been widely regarded as introducing charge-carrier traps that degrade mobility and stability. However, this view is based on conventional deoxygenation methods (annealing, sublimation) that cannot fully remove trace oxygen, leaving residual oxygen that complicates interpretation. Therefore, the true role of oxygen in OSCs requires re-investigation using an effective, nondestructive method capable of completely removing trace oxygen. The study aims to quantify inherent trace oxygen in a variety of OSCs, develop a reliable de-doping strategy, and elucidate how trace oxygen affects electronic structure, trap states, and charge transport, ultimately enabling precise modulation of device properties.

The paper revisits decades of literature where oxygen exposure was linked to trap formation and performance degradation in small-molecule and polymer OSCs (e.g., pentacene, rubrene), with effects observed via deoxygenation attempts (annealing, sublimation) and oxygen-related gap states reported by spectroscopy and theory. Prior studies documented oxygen-related traps impacting transistor characteristics, oxygen-enhanced photoconductivity, and gap states induced by oxygen or hydrogen. More broadly, extrinsic factors at ppm levels, including water and substrate silanols, have been shown to dominate transport by introducing traps or altering surface energetics. Despite this, the assumption that oxygen solely acts as a carrier trap persisted without acknowledging potential inherent oxygen dopants embedded within purified OSC lattices. The present work challenges this view by demonstrating that trace oxygen is inherent and stable, and acts as an acceptor that pre-empties donor-like traps, thereby explaining p-type tendencies and inhibited n-type behavior, and rationalizing perplexing observations such as Fermi level proximity to HOMO, unexpectedly high carrier densities/conductivities, low-temperature subthreshold swing deterioration, and sensitivity to ionization gauges.

- Materials and purification: Small-molecule OSCs (e.g., DNTT, C10-DNTT, PTCDI-C8, NDI-cy6, DHF-4T, PDI-CN2), conjugated polymers (P3HT, N2200), and other molecules (PEN, DPA, CuPc, TIPS-pentacene) were sourced commercially; small molecules were purified by triple physical vapor transport. High-purity metals used for electrodes. - Device fabrication: Predominantly bottom-gate/top-contact OFETs on highly doped Si/300 nm SiO2. Substrates O2-plasma treated then modified with OTS. Small molecules thermally evaporated (20–30 nm, 0.1 Å s−1) at 60–100 °C; PTCDI-C8 and other n-type molecules deposited on PS-coated SiO2. TIPS-pentacene by drop-casting on PS. Polymers spin-coated (10 mg mL−1 in chlorobenzene) and annealed. Au (typically 20 nm) used for S/D contacts; additional device structures included bottom-contact with Ag/PFBT and single-crystal devices (rubrene; C8-BTBT) prepared via specialized methods. Ultrathin (~5 nm) DNTT films prepared for rapid cycling demonstrations. - De-doping (deoxygenation) method: An elaborately controlled soft DC glow plasma (H2, N2, or Ar) operated typically below 20 W was used to remove inherent oxygen dopants without damaging OSC morphology/crystallinity (verified by XRD and AFM). Optimized treatments usually <1 min. The plasma, environmental control chamber, and glovebox were vacuum-interconnected to avoid air exposure. - Re-doping method: Exposure to O2 under illumination (Xe lamp up to 300 W or LED with selectable wavelengths 450–435 nm, 597–577 nm, 760–622 nm). Illumination accelerates oxygen re-doping; higher-energy light yields faster re-doping. Encapsulation (glass-glass) can suppress re-doping. - In-situ and quasi-in-situ characterization: UPS (He Iα, 21.22 eV) and TOF-SIMS connected via vacuum pipeline for sequential measurements pre- and post-plasma without air exposure. UPS used to monitor secondary electron cutoff and Fermi-level positions; TOF-SIMS quantified elemental profiles (O, C, S). EPR (X-band) tracked organic radical cations (ORCs) and detected superoxide anions via DMPO spin-trapping (DMPO-OOH). Spin density N_spin estimated via Curie law. Low-temperature EPR and transport used where applicable. - Electrical measurements: OFET transfer/output characteristics measured in the home-made system (Agilent B1500A, PDA FS380). Post-plasma measurements ~1 min after treatment; post-illumination ~3 min after (to avoid photocarrier effects). Threshold voltage (V_T), mobility (using large overdrive), subthreshold swing, hysteresis, and contact resistance extracted. DOS in the gap estimated using a classical FET method. Temperature-dependent transport analyzed within multiple trapping and release (MTR) and hopping frameworks. - Computation: First-principles molecular dynamics (FMD) simulations to locate oxygen within intralayer/interlayer sites; DFT to evaluate DOS and charge distribution with/without oxygen incorporation. - Parameter modulation demonstrations: Procedures to tune V_T to ~0 V (both from positive and negative starting values), enhance conductivity via re-doping (e.g., C10-DNTT to 725 S m−1), and invert transport polarity (TIPS-pentacene p→n via de-doping; N2200 n→p via re-doping).

- Inherent trace oxygen in OSCs: Even after triple purification, trace oxygen (~10^15 cm−3) is present in a wide range of OSCs (verified by TOF-SIMS and EPR). EPR of freshly prepared DNTT films shows an identifiable signal (g ≈ 2.0032–2.0034) attributed to DNTT radical cations (S = 1/2); DMPO spin-trap detects superoxide (DMPO-OOH). This indicates geminate superoxide anions (O2−) and organic radical cations (ORCs) as stable, inherent dopant pairs. FMD suggests oxygen resides in intra-/interlayer lattice sites, explaining stability and removal difficulty. - UPS and band alignment: UPS shows both secondary electron cutoff and Fermi level shift by ~0.35 eV towards higher energy after de-doping, moving EF away from the HOMO and evidencing successful removal of acceptor dopants. - Non-destructive de-doping: Soft plasma (H2/N2/Ar) under optimized conditions preserves film crystallinity/morphology (XRD, AFM unchanged) while reducing oxygen (TOF-SIMS) without altering C or S signals. - Trap pre-emptying by oxygen: DOS extraction shows that oxygen doping pre-empties donor-like traps, releasing EF pinning, reducing subthreshold swing, and enabling mobile hole carriers (p-type). After de-doping, donor-like traps become occupied, subthreshold swing deteriorates, contact resistance increases, and p-type conduction diminishes or disappears. - Transport regimes: Fresh DNTT OFETs follow MTR behavior; de-doped devices show hopping-dominated transport due to increased trap influence. Despite increased traps, intrinsic mobility decreases modestly (<50%) when re-extracted at higher gate overdrive, indicating mobility is largely a material property not heavily altered by doping. - Device-level impacts (p-type OSCs): Progressive negative V_T shifts and off-current reduction occur with increasing de-doping power; at strong de-doping, p-type transfer vanishes and conductivity becomes immeasurably low. - Reversibility: Re-doping under O2 with illumination rapidly and nearly fully restores the initial p-type behavior and EPR signals; cycling between de-doped and re-doped states is fast and reversible in ultrathin (~5 nm) films, with switching completed within minutes and multiple cycles achievable in ~20 min. - Enhanced n-type performance upon de-doping: Removing oxygen markedly improves electron transport. PTCDI-C8 mobility increases by >10× to ~2.2 cm^2 V^−1 s^−1 with improved V_T and Ion/Ioff; four additional n-type OSCs (N2200, NDI-cy6, DHF-4T, PDI-CN2) show similar mobility gains. This indicates oxygen dopants inhibit electron transport generally across OSCs. - Precise parameter modulation: V_T of diverse OFETs can be tuned close to 0 V without mobility loss. Conductivity of C10-DNTT increases by >100× to 725 S m^−1 upon re-doping, with mobility only slightly reduced (5.2→3.6 cm^2 V^−1 s^−1); the corresponding carrier density is ~2 × 10^18 cm^−3 (σ = neμ). Polarity conversion demonstrated: TIPS-pentacene (normally p-type) becomes n-type after de-doping and reverts to p-type after re-doping; N2200 (n-type) displays p-type behavior after re-doping, reversing with de-doping. - Mechanistic model: Oxygen behaves as an acceptor with states near the HOMO edge, reducing depletion width and Schottky barrier, facilitating injection (TE/TFE). De-doping shifts EF away from HOMO, broadens depletion region, increases injection barrier, decreases hole concentration, and reveals trap-limited transport.

The study overturns the prevailing assumption that oxygen in OSCs primarily forms carrier traps that degrade mobility. Instead, inherently present trace oxygen acts as a stable acceptor dopant forming O2−/ORC pairs that pre-empty donor-like traps, release EF pinning, and generate mobile hole carriers—explaining why many OSCs exhibit p-type behavior intrinsically and why n-type performance is commonly suppressed. By removing these acceptors via soft plasma de-doping, EF moves away from the HOMO, depletion width increases, and injection barriers grow, leading to diminished p-type conduction, increased subthreshold swing, and a shift from MTR to hopping transport. Conversely, re-doping under O2 illumination restores the acceptor states and device performance, with higher-energy light accelerating the process. The generalized observation across numerous OSCs indicates that trace oxygen is a ubiquitous, lattice-embedded dopant that has governed many reported device behaviors. This framework clarifies several previously perplexing phenomena, including EF alignment near the HOMO, unexpectedly high conductivities, low-temperature S deterioration, and sensitivity to vacuum instrumentation. Practically, the reversible de-doping/re-doping strategy offers a nondestructive, precise means to tune polarity, carrier density, conductivity, threshold voltage, and mobility, expanding the accessible property space and enabling better design of organic electronics.

This work introduces a nondestructive soft plasma deoxygenation (de-doping) method that removes inherent trace oxygen dopants from OSC lattices and reveals their true role: oxygen acts as an acceptor that pre-empties donor-like traps, underpinning the ubiquitous p-type character of many OSCs and suppressing n-type transport. The findings provide a coherent explanation for long-standing observations in organic electronics and establish a basis for reinterpreting charge transport in OSCs. The demonstrated reversible de-doping/re-doping under O2 illumination is green, facile, and efficient, enabling precise, cyclic modulation of key device parameters (polarity, conductivity, V_T, mobility) without structural damage. Future work may extend mechanistic understanding to polymers, where structural complexity likely introduces additional pathways and interactions, and optimize process conditions and encapsulation strategies for device stability and integration in manufacturing.

- Polymer systems: Although similar de-doping/re-doping behaviors were observed in polymers, the underlying mechanisms may be more complex and not fully consistent with small-molecule OSCs due to differences in microscopic and band structures, complicating generalization. - Film thickness and uniformity: Ultrathin (~5 nm) films were used to accelerate cycling, but such films often suffer from poor uniformity and stability; most results rely on regular thickness films, which may limit direct translation of ultrathin-film cycling speeds to practical devices. - Environmental stability: De-doped states can gradually re-dope upon oxygen exposure; stable maintenance of the de-doped state requires encapsulation, which may impose processing constraints. - Scope of materials: While a broad set of representative OSCs was tested, generalization to all polymers and device architectures may require further validation and optimization of plasma parameters to avoid subtle damage or side reactions.