Francis J.
Dent
a,
David
Harbottle
b,
Nicholas J.
Warren
b and
Sepideh
Khodaparast
*a
aSchool of Mechanical Engineering, University of Leeds, Leeds, LS2 9JT, UK. E-mail: s.khodaparast@leeds.ac.uk
bSchool of Chemical and Process Engineering, University of Leeds, Leeds, LS2 9JT, UK
First published on 14th March 2023
The breath figure (BF) method employs condensation droplets as dynamic templates for patterning polymer films. In the classical approach, dropwise condensation and film solidification are simultaneously induced through solvent evaporation, leading to empirically derived patterns with limited predictability of the final design. Here we use the temporally arrested BF methodology, controlling condensation and polymerisation independently to create diverse BF patterns with varied pore size, arrangement and distribution. External temperature control enables us to further investigate and exploit the inherent reversibility of the phase change process that governs the pattern formation. We modulate the level of subcooling and superheating to achieve subsequent regimes of condensation and evaporation, permitting in situ regulation of the droplet growth and shrinkage kinetics. The full reversibility of the phase change processes joined with active photopolymerisation in the current approach thus allows arresting of predictable BF kinetics at intermediate stages, thereby accessing patterns with varied pore size and spacing for unchanged material properties and environmental conditions. This simultaneous active control over both the kinetics of phase change and polymer solidification offers affordable routes for the fabrication of diverse predictable porous surfaces; manufacture of monolithic hierarchical BF patterns are ultimately facilitated through the advanced control of the BF assembly using the method presented here.
The BF method harnesses the growth and self-assembly of condensation droplets to induce honeycomb-patterns (HCP) on the surface of polymer films.10 This maskless, bottom-up surface patterning approach, herein described as the classical BF approach, uses long chain and end-functionalised polymers dissolved in a highly volatile solvent under the presence of a humid environment.11,12 Solvent evaporation reduces the temperature of the polymer solution, inducing condensation nucleation and growth at the air–polymer interface. Condensation droplets grow and self-assemble into a hexagonally packed array as the solvent evaporates completely, solidifying the templated polymer film prior to evaporation of the droplets themselves. While the droplet stabilisation and ordering mechanisms in classical BF remain the topic of continuous ongoing investigations, the competing interfacial phenomena of soluto-, thermo- and capillary instabilities, as well as aggregation of a polymer protective layer and solidification are known to suppress droplet coalescence, promoting self-assembly on mono- and multi-layered arrays of highly-packed pores.13–18 Total evaporation of the solvent is essential to the creation of the final porous structure, therefore, a specific output pore size and spacing is achieved for given experimental inputs of material properties and environmental conditions.
The facile nature of the classical BF approach lends itself to the passive coupling between the solvent evaporation and polymer curing that foresees both the initiation of condensation and termination of droplet growth with the curing of the pattern.17–20 While offering opportunities to create varied self-assembled patterns,21–23 this complex interconnection between the dynamic and non-equilibrium condensation/evaporation processes restricts the direct translation of the mechanistic knowledge of drop-wise condensation kinetics24 towards predictive design of pore size and spacing in the classical evaporation-driven BF21,25,26 approach. Substituting the passive evaporation-driven cooling of the polymer interface with external film cooling, we eliminate the coupling and enable in situ interrogation and modulation of the BF patterns through access to discrete intermediate designs within a single dynamic process.
The ability to form a stable BF array is generally governed by the chemical composition and concentration of the polymer/solvent.21,27–30 Therefore, full systematic control in the classical BF is fundamentally limited due to the stability constraints of the condensation droplets. If complete solvent evaporation and polymer solidification does not occur fast enough, the droplets will either tend towards coalescence31–33 or, given the right conditions, form multi-layer porous structures14,30,34,35 due to the sinking of the droplet array. The resultant pore size is influenced by the condensation induction time and the duration/level of subcooling below the dew point, ultimately dictated by the environmental conditions.36 Consequently, relatively narrow system-specific working windows are often empirically derived to achieve regular BF patterns at the end of the process, yielding limited systematic control over pore size without substantial modification of material or environmental conditions. Unlike the classical BF approaches where the focus has been on attaining HCP with monodisperse and non-coalescing droplets, in this work we aim to establish direct links with the physics of condensation and evaporation on liquid films to deterministically design monolayer BF patterns of predictable pore size and arrangement.
Recent attempts have been made to further advance the BF pattering approach towards modulating pore shape22,23,37 and generating hierarchical structures.38–40 The specific pore shape, which is primarily dictated by the interfacial tension at the three-phase contact line, has been altered through the use of non-aqeuous vapors,22,41 strict temperature/humidity regulation,23 and employing additives42 or sacrificial layers.43 Secondary patterns have been created via consecutive processes such as large-scale re-shaping of the pre-patterned substrates44 or BF patterning on physically confined 3D substrates.39,45–47 The use of combined techniques with nanoparticles48–51 or other self-assembly and phase separation techniques40,52–54 have also proven successful for the fabrication of hierarchical BF patterns. The reliance upon hybrid approaches, however, further narrows the range of operational working parameters and choice of materials, impairing the simplicity of the BF method with little to no deterministic control over the final design. Exploiting the reversibility of the condensation and evaporation regimes, we demonstrate that a single process can be implemented to attain monolithic hierarchical BF patterns with bimodal pore size distributions.
We recently described the adapted temporally arrested BF approach to control the BF pore size systematically using photopolymerisation and external cooling.55 Our main goal in this article is to identify control parameters that allow fabrication of temporally arrested BF patterns of predictable pore diameter and spacing through quantitative characterisation of the reversible phase change process. Although reversibility can be practised to some extent in the evaporation-driven breath figure approaches,23,56 full reversibility of the condensation process57 and access to intermediate patterns are only possible in methods that eliminate solvent evaporation and provide active polymerisation. Here, we explore the possibility of inducing and retracting the reversible BF pattern via external temperature control, implementing subsequent regimes of subcooling and superheating with the goal to achieve in situ tuning of the size and spatial arrangement of BF patterns, as schematically illustrated in Fig. 1. Further, we investigate the feasibility of restarting the BF process to initiate a second condensation regime, leading to monolithic, multi-scale patterns following a single fabrication route.
The subcooling temperature was set based on the ambient laboratory temperature (T0) and relative humidity (RH). The subcooling level ΔTC = |TDP − TP| was defined as the difference between the set temperature on the Peltier device TP and the dew point temperature TDP calculated by measuring the ambient conditions before each experiment. Similarly, evaporation of the BF pattern was controlled at a constant superheating level, ΔTH, with TP > TDP. Supersaturation temperatures between 0 °C to 10 °C prevented freezing (during condensation), while enabling relatively slow kinetics that could be captured by our imaging system and rendered the UV curing time relatively negligible. All reported subcooling and superheating temperatures are reported with respect to the set temperature of the Peltier device. A maximum ±2.5 °C temperature difference between the Peltier surface and glass coverslip was found during the experiments, with negligible temporal-temperature lag during cooling/heating ramps.
To create porous solid surfaces, the liquid BF patterns were arrested and cured at discrete times with a 5 s exposure of 365 nm UV irradiation using a 425 mW cm−2 collimated light source (Dymax Redicure550) at 25 mm distance from the substrate.
The average interdroplet spacing, Lc, was analysed in different frames captured during the evaporating regime by calculating the average euclidean distance between droplets sharing a vertex in a corresponding Voronoi diagram (Fig. S3, ESI†). Parameters were calculated for every recorded frame and averaged across 10 s intervals to reduce the effects of thresholding errors in individual frames.
Under optimised material and environmental conditions, stabilised condensation droplets in classical BF grow mainly by diffusion, forming monodisperse HCP with negligible coalescence between droplets.26 In contrast, the condensation droplets in our BF patterns appear at high initial nucleation density and grow due to diffusion and significant coalescent events. Fig. 2a shows the kinetics of droplets’ growth on the NOA63 photocurable polymer film which was previously found to promote high-density nucleation.55 Deterministic control of the droplet size in the BF patterns is achieved by maintaining a constant subcooling level at temperatures below the dew point. By permitting droplet coalescence, high packing can be maintained within a single layer BF pattern with no upper limit for the droplet size.14 Although disturbing the hexagonal packing order and inducing polydispersity, the coalescent-driven growth allows systematic control of pore diameter in a repeatable and predictable manner.
The monotonic growth rate shows conformation to the linear trendline of D ∝ t with visible highly packed and self-similar condensation growth (Fig. 2a). In this coalescent-dominated regime, the area fraction of condensation droplets remains constant at Af = 0.71 ± 0.05, demonstrating near optimal packing when accounting for the slight disorder due to the droplet polydispersity and the thin viscous polymer film stabilising the droplets, as observed in Fig. 2b.59 While variation in the droplet diameters occurs due to coalescence between neighbouring droplets, the polydispersity throughout the analysis in this regime remains low and near constant at g = 0.10 ± 0.05. This range is similar to that found in original studies on non-stabilised BF droplet growth on liquid films,58 whilst being within the ranges reported in non-HCP21,25,60,61 BF patterns and late-time condensation regimes.62 The self-similar dynamic BF pattern on the liquid polymer film (Fig. 2b) can be arrested via UV irradiation at any moment of time to achieve a patterned surface of known pore diameter and area coverage. Transitioning from droplets in liquid films to pores in the cured polymer, there is no significant change in the shape across the experimental tests or after UV exposure due to the largely dominant interfacial forces at these length scales.
Using our real-time optical microscopy analysis, we characterise the evaporation kinetics of droplets within BF patterns formed on NOA63 at constant superheating levels. While in the highly packed condensation regime, droplets are being pushed by the neighbours and rearrange due to coalescence, they remain fairly stationary in the evaporation regime indicating negligible capillary interactions or droplet diffusion in the length and time scales studied here. During evaporation, the radially shrinking droplets maintain a constant interdroplet spacing, Lc ≈ 1.2 × DL0, with DL0 the maximum droplet diameter attained in the highly packed condensation regime (Fig. 3a). Unlike in the condensation regime where the area fraction covered by the BF pattern remains unchanged, evaporation leads to continuous reduction in the droplet diameters whilst the distance between them is maintained; the area fraction can thus be significantly reduced in a controlled manner during the evaporation regime by understanding the evaporation kinetics. In the example shown in Fig. 3, the area fraction covered by the BF pattern decreases by more than 85% from the initial value of Af = 0.68.
In cured BF patterns, while pore size modulation at constant high surface packing was achieved via temporally arrested BF, the evaporation step allows diameter modulation of the equally spaced pores, reducing the area fraction.
Droplet diameter continuously decreases during the evaporation, suggesting that evaporation of BF patterns on a liquid substrate occurs in a constant contact angle regime, where the droplets maintain a spherical cap with diminishing drop height and contact area. Previous theoretical and experimental investigations confirmed that the mass, M, of a sessile droplet evaporating on hydrophobic substrates in isothermal conditions decreases according to M2/3 ∝ t. The relationship is derived in the regime where the contact line recedes without significant change in the contact angle.63,64 Relating the droplet mass to its diameter D predicts D2 to decrease linearly with time during the evaporation if the droplet shapes remain self-similar. To further investigate the viability of the combined condensation and evaporation cycles for pattern modulation and the validity of the constant contact angle regime for droplet evaporation, analysis of the time-dependent evolution of evaporating BF patterning droplets for a range of operating and environmental conditions is performed. In order to eliminate any smearing effect caused by size polydispersity (Fig. S5, ESI†) and gain a better understanding of the evaporation mechanism, individual droplets are tracked and analysed in Fig. 4, see more details in the ESI.† All data is presented in square diameter to allow comparison with the available prediction models.
![]() | ||
Fig. 4 Kinetics of BF evaporation analysed for individual droplets and averaged over 10 s intervals, see Fig. S5 in the ESI† for more details. (a) The square of droplet diameter monotonically decreases with time. The initial average diameter in the fully packed BF is controlled by varying the duration of the condensation regime tc, before starting the superheating cycle. Data corresponds to ΔTH = 5 °C, T0 = 19 °C and RH = 66%. Snapshot images show patterns with similar average square diameter DL2 = 30 μm2. (b) Higher levels of superheating speeds up the BF pattern evaporation. Snapshots show patterns with similar square normalised diameters (DL/DL0)2 = 0.7 and area fractions, obtained at different superheating levels. Data was captured at T0 = 22 °C, RH = 55% and the respective ΔTH detailed on the graph. (c) Normalising pattern diameters and time allows the collapsing of all data presented in (a) and (b). SEM images correspond to the morphology of a typical BF pore (top), and sinking pores observed at late-stage of the evaporation cycle (middle and bottom). |
For a given ΔTC and RH, Lc is only a function of the time from the start of the BF formation in the highly packed condensation process, tC. Therefore, combining the condensation and the evaporation steps provides a powerful handle for dynamic spatial modulation of BF patterns where both droplet diameter and area fraction can be controlled. Initiating the evaporation at different times from the start of the condensation nucleation allows the generation of BF patterns with different diameters and interdroplet spacing for given subcooling and superheating conditions. Fig. 4a shows diverse patterns with similar average droplet diameters but varying interdroplet spacing obtained by tuning tC. The snapshot images correspond to DL2 = 30 μm2, showing that BF patterns can be arrested with different arrangements and area coverage at constant pore diameters. Additional control over the evaporating BF pattern is obtained by adjusting the level of superheating set by the value of ΔTH, i.e. higher superheating levels yield faster evaporation and smaller final diameters with larger interdroplet spacing at a given time from the start of the evaporation cycle. Constant average pattern diameter and interdroplet spacing can thus be obtained by arresting varying superheated evaporation cycles at different times, creating effectively identical patterns using different input parameters, see example snapshots of patterns in Fig. 4b.
While the trends in droplet diameter reduction appear to be similar in Fig. 4b, higher levels of superheating yield faster shrinking of the droplet. Normalising the x-axis by an arbitrary time, t0.4, at which the average droplet diameter reaches 40% of its initial value (DL/DL0 = 0.4), accounts for the impact of superheating and in turn collapses all data captured at varied operational and environmental conditions onto a single curve (Fig. 4c). A linear trend, (DL/DL0)2 ∝ t/t0.4, provides a fair prediction of the collapsed data until (DL/DL0)2 = 0.5 is reached. At later times we observe a significant increase in the droplet shrinkage rate with deviation from the predicted linear trend (Fig. 4c). At these times, some of the condensation droplets in the field of view appear out of focus, suggesting an increasing frequency of droplets moving downward relative to the top interface of the polymer film.65 This vertical translation is clearly observed in the SEM images of the cured films’ cross sections presented in Fig. 4c, where droplets sink below the interface and eventually become fully encapsulated within the polymer, see further details in Fig. S6 in the ESI.† As we employ the BF approach for surface patterning, later evaporation times where initial water droplets are no longer all on the surface are not considered for further analysis. Compared to the condensation regime, the polydispersity in droplet diameters gradually increases to g = 0.20 ± 0.05, up to a maximum value of g ≈ 0.3 at the final stages of evaporation where the downward translation of droplets is observed. This increase arises from the initial polydispersity in the highly packed droplet population, augmented by the different evaporation rates and vertical translation of droplets within the field, especially at smaller diameters (Fig. 4c).
Fig. 5 schematically illustrates the experimental procedure and real-time microscopy images of the BF patterns produced through a simple combined cycle of condensation, evaporation and re-condensation: (1) an initial subcooling period of condensation is completed to establish highly packed BF patterns. This regime dictates the specific interdroplet spacing for the large droplet domains as previously elucidated. (2) The system is then superheated to reduce the droplet packing as the droplet diameters decrease during the evaporation regime. (3) Restarting the subcooling regime for a second time, we initiate the re-condensation regime manifested by the appearance of new condensation nucleation sites on the free surface of the polymer, as well as the growth of the original droplets.
SEM images of the cured films and the corresponding histograms of pore diameters are presented in Fig. 5(ii and iii), respectively. The visible diameters on the liquid (i) and solid (ii) films differ as they are associated with the full droplet diameter visible during liquid analysis and the pore opening diameter visible with SEM (Fig. S4, ESI†). The pattern histograms (iii) show the polydispersity in pore diameters and their distribution at different times. The active regulation of the substrate temperature allows us to control the interdroplet spacing in the primary regime and ultimately modulate the area fraction to create cured porous surfaces of desired characteristics. Through consecutive regimes of cooling and heating, we harness the predictable kinetics and reversibility of the temporally arrested BF patterning approach to achieve multiscale structures. Optical microscopy video of a continuous condensation, evaporation and re-condensation can be viewed in the ESI.†
The current temporally arrested BF method provides dynamic, in situ control over the size and spatial distribution of interfacial pores in polymer films. We explored the inherent reversibility of the phase change mechanism governing the process prior to curing of the polymer. Condensation and evaporation kinetics on the liquid films were analysed in controlled regimes of subcooling and superheating, respectively, modulating the average droplet size at constant surface coverage before modulation of the surface coverage. Ultimately, hierarchical and multiscale patterns were formed through combined condensation–evaporation cycles.
An example catalogue of the cured designs is presented in Fig. 6; diverse patterns are created with constant packing but an order of magnitude differing diameters (Fig. 6a), constant interdroplet spacing at varying diameters (Fig. 6b), constant diameters at differing area fractions (Fig. 6c), and finally bimodal distributions of pore diameters (Fig. 6d). The full BF reversibility for surface patterning practised here is a unique feature of the temporally arrested approach thanks to the external temperature control, rapid UV curing and the elimination of solvent evaporation. Further control over the BF pattern and pore morphologies can be achieved through tailoring properties of the photocurable polymer66 and finely tuning the environmental parameters23 to facilitate viable commercial and industrial exploitation of the method.
![]() | ||
Fig. 6 SEM images of the cured patterns attainable from the temporally arrested BF process, showing the diversity of the programmed design. (a) Constant area fraction at varying diameters. (b) Constant interdroplet spacing at varying diameters. (c) Constant diameters at varying interdroplet spacing. (d) Varied hierarchical designs. Operational parameters are included in Table S1 in the ESI.† |
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d2sm01650h |
This journal is © The Royal Society of Chemistry 2023 |