Strategies to control humidity sensitivity of azobenzene isomerisation kinetics in polymer thin films

The study addresses how to control and optimize the humidity sensitivity of azobenzene thermal cis–trans isomerisation kinetics in polymer thin films for optical humidity sensing. Azobenzenes are widely used molecular photoswitches whose switching properties can be tuned by substitution and environment. Prior work showed hydroxyazobenzenes exhibit strong environmental dependence of isomerisation rates, particularly via hydrogen bond-assisted azo–hydrazone tautomerism. A single prior sensing example based on isomerisation kinetics used hydroxyazobenzenes in polymer films for optical humidity sensing. However, key questions remained regarding how sensitivity depends on azobenzene structure, polymer matrix water uptake, generality beyond hydroxyazobenzenes, and long-term reliability for commercialization. This work systematically investigates these factors, aiming to establish design principles for practical, accurate, and robust humidity sensors.

• Photoswitch performance depends on switching wavelengths, state stability, solubility, and photostability; application-specific requirements vary across biomedicine, energy storage, electronics, and catalysis.
• Azobenzenes are prominent due to large conformational changes upon isomerisation and tunability via ring substitution; ortho-substitution enables visible-light bidirectional switching; protonation can push switching into NIR; cis stability can be tuned from milliseconds to years, including via heteroaromatic azo-compounds.
• Environmental factors (temperature, polarity, concentration, pressure) substantially affect azobenzene isomerisation rates; hydroxyazobenzenes show orders-of-magnitude differences in polar vs nonpolar media.
• In sensing, azobenzenes have mostly been used as colorimetric probes based on spectral changes on analyte binding. The only prior example of sensing based on isomerisation kinetics is optical humidity sensing using hydroxyazobenzenes in polymer films (Poutanen et al., 2018), attributed to hydrogen bond-assisted azo–hydrazone tautomerism.
• Knowledge gaps: dependence of sensitivity on azobenzene chemical structure and polymer matrix, generality beyond hydroxyazobenzenes, and sensor reliability for practical deployment.

Overall approach: Establish and analyze humidity-dependent thermal cis–trans isomerisation kinetics of various tautomerizable azobenzenes embedded in polymer thin films. Correlate kinetics with polymer water absorption behavior and azobenzene content. Support mechanistic insights with DFT calculations. Demonstrate device-level feasibility with concept prototypes.

Sensing concept and kinetic model:

• Monitor azobenzene absorbance changes after photoexcitation due to distinct trans/cis spectra. In solids, kinetics are better described by a stretched exponential (Kohlrausch–Williams–Watts) model. Report a single effective rate constant k and stretching parameter β.
• Empirical humidity dependence modeled as k(RH) = k0 · e^(μ RH), where μ (also denoted v in parts of the text) quantifies humidity sensitivity at a given temperature.

Materials:

• Azobenzenes: hydroxyazobenzenes (2PAP, CN-OH, OH-OH), heteroaryl hydroxyazobenzenes (Pyr-OH, AIZ-OH), aminoazobenzenes (NH2-NH2, NO2-NH2, DMA-NH2), and dimethylamino derivatives with varying para substituents (DMA-OH, DMA-COOH, DMA-m-COOH). Non-tautomerizable control: m-OH-m-OH.
• Polymers: P4VP (hygroscopic, H-bond accepting), PMMA (weakly hygroscopic, near-linear water isotherm), EC (moderately hygroscopic, superlinear water isotherm). Polystyrene was unsuitable due to poor miscibility.

Synthesis and procurement:

• Several azobenzenes purchased (purities given). Synthesized compounds: AIZ-OH, m-OH-m-OH, DMA-m-COOH, Pyr-OH, 2PAP via literature/adapted routes; structures verified by NMR and MS.
• Polymers (P4VP, PMMA, EC) sourced commercially.

Film fabrication:

• Polymer and azobenzene dissolved in ethanol, ethyl acetate, DMF, or mixtures per target molar ratios (typically 1:8 azo:polymer repeat units; EC films set to 1 azo per 8 –OH groups). Spin-coating onto glass at 1000 rpm for 40 s; drying at 60 °C for 5–10 min; DMF films vacuum-dried at 40 °C overnight. Film thickness ~150–300 nm by AFM.

Optical measurements:

• Absorption spectra measured post-fabrication. Thermal back-relaxation measured in a humidity-controlled chamber (23 °C unless stated) using DH-2000 BAL monitoring light and fiber spectrometer. Short 100 ms LED pulses (wavelength slightly longer than π–π* λmax) used for photoexcitation. Neutral density and band-pass filters applied as needed. Kinetics analyzed with stretched exponential model to extract k and β.
• Solution lifetimes measured in dry THF (and THF with water for CN-OH) using a UV–vis spectrophotometer and LED source.

Water absorption measurements:

• QCM-D with humidity module at 23 °C to determine polymer and composite film water uptake isotherms (reported as wt% water). Humidity controlled via saturated salt solutions. Dissipation changes monitored to infer mechanical changes.
• DVS measurements (20–23 °C) corroborated QCM-D isotherms; required scraping multiple films to achieve mg-scale mass.

Device demonstrations:

• First-generation concept: bulky unit with LED and photodiode measuring isomerisation kinetics to infer ambient RH; continuous monitoring over 6 weeks.
• Second-generation handheld device: multi-wavelength excitation, sticker-type sensor arrays with different sensing films; used to monitor drying of fresh concrete over 11 weeks (100%→~85% RH).

Computational studies:

• DFT (ωB97X/def2-TZVP) with CPCM(THF) to compute energies of cis-azo vs cis-hydrazone tautomers and transition states for rotation (hydrazone) and inversion pathways. Frequency analyses confirmed minima/TS. Energetic trends correlated with observed kinetics and humidity sensitivity.

• Humidity sensitivity requires tautomerizable structures: • Non-tautomerizable m-OH-m-OH shows extremely slow back-isomerisation (cis lifetime ~5 h in P4VP at 22 °C) with no humidity dependence, whereas tautomerizable OH-OH shows fast, strongly humidity-dependent kinetics.
• Hydroxyazobenzenes (para-substituted and heteroaryl): all display strong humidity sensitivity (4–6 orders of magnitude dynamic range). Para substituent modulates absolute rate at given RH; strong electron donors (e.g., DMA) slow and strong acceptors (e.g., CN) accelerate kinetics. Heteroaryl Pyr-OH is notably fast with slightly weaker sensitivity.
• Aminoazobenzenes: display weaker humidity sensitivity (≤~2 orders of magnitude). D–A configuration (NO2–NH2) increases rate but not sensitivity compared to hydroxy analogs.
• Carboxylic acid substitution on dimethylaminoazobenzenes (DMA-COOH, DMA-m-COOH) yields humidity sensitivity comparable to or exceeding hydroxy substitution and is largely independent of COOH position (para vs meta).
• Representative quantitative results at 22 °C, films in P4VP at 1:8 molar ratio (from Table 1): • 2PAP: τ50%RH = 59.5 s; μ = 0.1423 %−1; dynamic τ range ≈ 47 ms – 19.7 h. • CN-OH: τ50%RH = 9.1 s; μ = 0.1282 %−1; dynamic τ range ≈ 15 ms – 1.5 h (THF τ ≈ 1.8 h). • DMA-OH: τ50%RH = 125 s; μ = 0.1375 %−1; dynamic τ range ≈ 124 ms – 32.3 h (THF τ ≈ 48.3 min). • OH-OH: τ50%RH = 21.5 s; μ = 0.1332 %−1; dynamic τ range ≈ 27 ms – 4.6 h (THF τ ≈ 2 h). • AIZ-OH: τ50%RH = 34 s; μ = 0.1445 %−1; dynamic τ range ≈ 25 ms – 13.2 h (THF τ ≈ 24.4 h). • Pyr-OH: τ50%RH = 0.67 s; μ = 0.0985 %−1; dynamic τ range ≈ 6 ms – 103 s (THF τ ≈ 48 s). • NO2–NH2: τ50%RH = 0.48 s; μ = 0.0474 %−1; dynamic τ range ≈ 53 ms – 5.9 s (THF τ ≈ 1.5 s). • NH2–NH2: τ50%RH = 234 s; μ = 0.0382 %−1; dynamic τ range ≈ 33 s – 25.6 min (THF τ ≈ 13.6 min). • DMA–NH2: τ50%RH = 179.8 s; μ = 0.0360 %−1; dynamic τ range ≈ 24 s – 14.9 min. • DMA–COOH: τ50%RH = 69 s; μ = 0.1549 %−1; dynamic τ range ≈ 27 ms – 40.3 h (THF τ ≈ 28.6 min). • DMA–m–COOH: τ50%RH = 100 s; μ = 0.1519 %−1; dynamic τ range ≈ 48 ms – 52.4 h (THF τ ≈ 22 min).
• Polymer matrix effect (CN-OH at 1:8): • Water uptake at 94% RH: PMMA ~2 wt%, EC ~6 wt%, P4VP ~23 wt% (23 °C). PMMA is near-linear isotherm; EC superlinear; P4VP near-linear up to ~75% RH then accelerates. • Kinetics correlate with polymer water isotherms: PMMA yields lower rates and weaker humidity dependence; EC shows faster-than-exponential dependence mirroring its superlinear uptake; P4VP shows near-exponential k(RH) across the studied RH range.
• Azobenzene content effect in P4VP (CN-OH at 1:8, 1:4, 1:1 azo:repeat units ≈ 21, 35, 67 wt% CN-OH): • Water uptake at 94% RH decreases strongly with higher azo content (P4VP: 22.7 wt%; 1:8: 15.6 wt%; 1:4: 8.5 wt%; 1:1: 2.9 wt%). • Humidity sensitivity μ decreases with higher azo content: 1:8 μ=0.1282; 1:4 μ=0.1213; 1:1 μ=0.0825. 1:1 shows faster rates at low RH (likely due to intermolecular interactions) but reduced sensitivity overall.
• Mechanistic insight from DFT: • 4-hydroxyazobenzenes: small cis-azo vs cis-hydrazone energy gaps (~1–2.5 kcal/mol) enable facile tautomerization; 4-aminoazobenzenes show much larger gaps (~9.9–13.2 kcal/mol), reducing tautomerization. • Hydrazone rotational TS barriers are low (≈1.3–7.4 kcal/mol) and correlate with faster isomerisation; inversion barriers separate molecules into fast (CN-OH, Pyr-OH, NO2–NH2, ~20.7–22.7 kcal/mol) vs slower (others, ~29.1–32.0 kcal/mol), matching observed base rates. • Proposed mechanism: water-promoted proton transfer to azo nitrogen shifts equilibrium toward hydrazone, enabling fast rotation-dominated isomerisation. This rationalizes strong humidity effects and COOH-substituted DMA compounds’ high sensitivity.
• Device demonstrations: • First-generation prototype continuously measured ambient RH for 6 weeks with stable operation. • Second-generation handheld, multi-wavelength device with sticker sensors monitored drying of concrete over 11 weeks; accuracy within ±5% RH vs reference, though inter-device discrepancies appeared; sensors remained robust despite harsh VOC-rich conditions. • Practical cis lifetimes for sensing considered ~10 ms–10 s to balance speed and simple optics; current materials cover parts of this range depending on RH and composition.

The findings directly address how azobenzene structure, polymer matrix, and azobenzene content control humidity-sensitive isomerisation kinetics. Tautomerizable azobenzenes, especially 4-hydroxy derivatives and dimethylamino azobenzenes bearing COOH groups, exhibit the strongest humidity sensitivity due to water-promoted azo–hydrazone tautomerization that favors a low-barrier rotational pathway. DFT corroborates that facile tautomerization combined with relatively high inversion barriers enhances the humidity dependence while allowing control over absolute rates via substituent electronics and heteroaromaticity. The polymer matrix mediates the interaction between azobenzene and environmental water. Strong correlations between polymer water uptake isotherms and kinetic dependencies show that linear uptake tends to yield exponential k(RH), whereas superlinear uptake introduces departures from simple exponential behavior. Thus, selecting polymers with appropriate water absorption (magnitude and isotherm shape) tunes both sensitivity and usable RH range. Azobenzene loading offers a fine-tuning lever: higher loading reduces water uptake (lower μ) yet can increase rate at low RH due to additional internal H-bond donors (–OH) and intermolecular interactions; however, it risks phase separation and stability issues. Together, these results provide a clear materials-by-design framework to tailor sensing performance (accuracy, speed, RH range, longevity) and demonstrate feasibility at device level, while highlighting areas where further optimization (e.g., measurement speed uniformity, accuracy) is required for commercialization.

This work establishes three practical strategies to control humidity sensitivity of azobenzene isomerisation kinetics in polymer thin films: (1) azobenzene molecular design (favor tautomerizable structures with appropriate electronic substitution to balance base rate and humidity response), (2) polymer matrix selection (choose materials with suitable magnitude and shape of water uptake isotherms and strong, stabilizing interactions with the azo dye), and (3) azobenzene content optimization (trade off sensitivity and speed against film stability and phase separation). Hydroxy- and COOH-substituted dimethylamino azobenzenes deliver strong humidity sensitivity, while aminoazobenzenes provide faster baseline rates with lower sensitivity. Polymers like P4VP, with higher water uptake and near-linear isotherms across the operating RH, support exponential sensitivity and wide dynamic ranges; EC’s superlinear uptake introduces non-exponential behavior that can be exploited or mitigated. Adjusting azobenzene content allows further fine-tuning but must avoid phase separation to maintain longevity. Future research should: (i) explore polymers with even faster, reversible water sorption and minimal hysteresis; (ii) map kinetics over full 0–100% RH ranges to better capture deviations from exponential behavior; (iii) expand molecular libraries emphasizing hydrazone-rotation pathways with controlled inversion barriers; (iv) optimize device optics/electronics and analysis algorithms to improve accuracy and standardize inter-device performance; and (v) study long-term stability and phase behavior under cycling and harsh environments.

• Experimental RH range for P4VP kinetics was practically limited to <80% RH, preventing thorough characterization of the superlinear uptake regime and potentially masking deviations from simple exponential k(RH).
• Linear fits of ln k vs RH are used for comparison but may not extrapolate accurately to extreme RH; deviations suggesting faster-than-exponential dependence were observed for some compounds.
• Measurement accuracy of the prototype devices remains limited (±5% RH vs reference) with noticeable inter-device discrepancies; measurement speed varies drastically with RH, constraining operational range.
• High azobenzene loading reduces sensitivity and increases risk of phase separation, compromising long-term film stability; balancing loading for performance vs longevity is nontrivial.
• DVS required pooling multiple films, which may introduce sample handling variability; QCM-D dissipation changes indicate mechanical property shifts at high RH that could influence kinetics and were not fully deconvoluted.