During the past few decades, cholesteric liquid crystals (CLCs) with intrinsic helical configuration of molecular directors have great perspectives towards a wide range of advanced photonic applications such as brightness-enhancement devices of liquid crystal (LC) displays, diffractive optical elements, smart windows, mirrorless lasers, and sensors^[1–5]. Thanks to the Bragg reflection of CLCs, which confers nontrivial optical functionalities, significant structural color can be generated to reflect circularly polarized (CP) light with identical handedness in the visible spectrum. In general, the colors of CLCs, exhibited by the selective reflection wavelength, depend on the helical pitch length ($p$) or/and the average refractive index (average RI, $n_{\mathrm{av}}$)^[6–8] in response to the external stimuli, such as temperature, electric field, and light irradiation. Also, some organic compounds can be incorporated into CLCs to alter the optical properties directly. The molecular structure of CLCs can be further modified with recognition fragments and utilized to absorb specific analytes. The uptake of specific analytes will affect $p$ and $n_{\mathrm{av}}$ and reveal different reflected colors consequently^[9–11]. This feature is beneficial for the detection of volatile samples such as alcohol, amine, and acetone^[12–14], providing a versatile sensing platform with several advantages like low cost, being power-free, and naked-eye detection.

However, two main limitations still exist in low molar mass CLCs and should be overcome in colorimetric methods, that is, the durability problem caused by fluidity, which makes them difficult to be prepared as portable systems, and the temperature-sensitivity problem, which may lead to a false-positive response. To address the above-mentioned issues, studies on recording and stabilizing the helical arrangement through the employment of mechanically robust polymer networks have attracted extensive attentions^[15–18]. Reactive mesogens (RMs) can passively “freeze” the anisotropic phase in the polymerization process to form a rigid porous structure. Notably, a number of pioneering works have exploited the concept of adopting cholesteric scaffolds as polymer templates to imprint supramolecular helical structures based on the multi-step “wash-out/refill” strategy^[19–21]. Novel cholesteric composites have been achieved by refilling different organic compounds, e.g., nematic and cholesteric mixtures, photoactive nematic copolymer, or optical adhesive. Very recently, Zheng et al. evaluated the interaction between the solvent and the polymer network by monitoring the shift of spectral reflection bands, providing a remarkable understanding of the micro-environmental chemical physics effect^[22].

In this work, we report on the optical response of polymerized CLC (PCLC) networks templated by the “wash-out/refill” method in the presence of organic compounds. The dynamic coloration was facilitated by two key approaches to diffuse organic compounds into the polymerized cholesteric networks. The first one is based on the alternative injection of two mutually soluble fluids, that is, a nematic LC (NLC) E7 and an organic solvent benzyl alcohol (BA), into a microfluidic channel and refilling the cholesteric scaffold integrated within in turn, therefore enabling real-time and reversible tunability. The second one is to explore the diffusion and interaction between PCLC networks and representative volatile organic compounds (VOCs, using alcohol as a model compound) with low concentration. This work is expected to extend the study of PCLCs as a dynamically tunable optofluidic reflector, visually readable sensor, or compact anti-counterfeit label in response to organic compounds.

The LC/monomer premixture was composed of E7 (71.8% mass fraction), R5011 (2.2%), RMs (25%), and photoinitiator Irgacure 651 (1%). RMs were made up of RM257, RM82, RM006, RM021, and RM010, with a weight ratio of 3:2:2:2:1^[23]. Two indium tin oxide (ITO) glass substrates were spin-coated with SD1 to form photoalignment layers before being bonded together using a 20-µm-thickness double-sided adhesive tape. Then, the LC cell was irradiated by linearly polarized visible light (LPVL, 405 nm, $50{\mathrm{mW}/\mathrm{cm}}^2$, 50 min) [Fig. 1(a-i)]. The premixture flew into the LC cell by means of capillary force at a temperature above the clear point (90°C). After being cooled down to room temperature slowly, both the reactive and nonreactive monomers were aligned in the planar state and showed standing helical structures [Fig. 1(a-ii)]. After the exposure under UV light (365 nm, $70{\mathrm{mW}/\mathrm{cm}}^2$, 10 min), the acrylate end groups of RMs were photopolymerized to form polymer network skeletons. As shown in Figs. 1(a-iii) and 1(a-iv), the unreacted RMs and E7 molecules also acting as pore-forming additives were removed in the following “wash-out” step by the permeation and exchange of organic solvents, and the cholesteric scaffolds can be readily refilled.

Elastic polydimethylsiloxane (PDMS) is an excellent choice for preserving optical properties of PCLC networks in microfluidic operation. Photoresist SU-8 was used in the design and fabrication of microchannels by soft lithography. The microfluidic device consists of two injection ports, a serpentine channel for fluid mixing, and a rectangular region of $20\mathrm{mm}\times 3\mathrm{mm}\times 0.1\mathrm{mm}$ for containing PCLCs [Fig. 1(a-v)]. Several meticulous manipulations were required in the integration of PCLC films into microfluidic channels. The PCLC film that adhered firmly to the substrate close to UV exposure was washed thoroughly in ethanol, reshaped carefully by a blade to fit in the rectangular region, and bonded immediately after oxygen plasma treatment.

In the microfluidic experiments, E7 ($n_{\mathrm{av}}=1.64$) flowed in the microchannel at a controlled flow rate by using a syringe pump, infiltrating the cholesteric scaffold, which was washed and refilled with BA ($n=1.54$) beforehand [Fig. 1(a-vi)]. BA was chosen because of its miscibility with E7, higher boiling point, and weaker volatility compared with other commonly used organic solvents like acetone, alcohol, etc. However, the mixing time should be prolonged because of a higher viscosity coefficient. As shown in Fig. 1(b), although hand shaking will accelerate the dissolution process, E7 and BA are initially separate in two layers due to slow diffusion. According to the Einstein–Smoluchowski equation and the Stokes–Einstein equation, the mutual diffusion distance $\mathrm{\Delta }x$ of E7 and BA molecules could be calculated as^[24]$$\langle {(\mathrm{\Delta }x)}^2\rangle =2D·t,$$(1)$$\begin{gathered} D=k_{i}T/6\pi {\eta }_{i}a \\ =C/{\eta }_i, \end{gathered}$$(2)where $D$ is the diffusion coefficient, and $t$ is diffusion time. The constant $C$ includes Boltzmann constant $k_i$, absolute temperature $T$, and molecular size $\alpha$. Thus, it can be seen that the diffusion coefficient $D$ is negatively correlated with the viscosity coefficient ${\eta }_i$. As shown in Fig. 1(c), by the injection of E7 with high ${\eta }_i$, BA molecules in the gap between the channel wall and the PCLC film were completely pushed out in a stratified form due to the limited $\mathrm{\Delta }x$.

Figure 2 shows the reflectance spectra measured at different fabrication stages by using a fiber spectrometer (Ocean Optics USB4000). The blue-shift of the reflection band occurred after the polymerization because of the shrinkage of film thickness, and the resulting contraction of helical pitch of the PCLC network formed on one single substrate as the tethering force from the other substrate was absent^[15]. Actually, the concentration of RMs was optimized to avoid the severe collapse of the helical polymer network when being fully dried to a sponge-like structure with dense nanopores. For a perpendicular incidence, the central wavelength of Bragg reflection ($\lambda$) and the reflection bandwidth ($\mathrm{\Delta }\lambda$) of CLCs are the following: $$\lambda =n_{\mathrm{av}}·p,$$(3)$$\mathrm{\Delta }\lambda =\mathrm{\Delta }n·p=(n_e-n_o)·p,$$(4)where $n_{\mathrm{av}}$ is the average RI, $p$ is the helical pitch, and $\mathrm{\Delta }n$ is the anisotropy of RI, which can be calculated by subtracting ordinary RI ($n_o$) from extraordinary RI ($n_e$). The substitution of organic compounds not only influences the overall $n_{\mathrm{av}}$ but also $p$ in terms of different swelling/deswelling effects of the CLC helical structure due to the differences in molecular affinity when incorporated in the porous cholesteric scaffold. Once NLC E7 infiltrated the dried film, the CLC film did not swell to its initial state due to a balance between elastic recovery and unexpected loss of unreacted RMs that are removed. This interpreted the phenomena where the reflection band cannot return to the initial wavelength position before the “wash-out” step. The inset shows the micrographs observed using a metallographic microscope (Nikon CM-70 L) at different fabrication stages.

By injecting E7 and BA alternatively, $\lambda$ can be varied between $\sim 450\mathrm{nm}$ and $\sim 600\mathrm{nm}$ reversibly, covering the chromaticity space from blue to orange, in less than 2 min for the flow velocity of 20 µL/min [Figs. 3(a) and 3(b)]. The reflection intensity is increased by injecting E7 with higher RI and $\mathrm{\Delta }n$. Moreover, as shown in Figs. 3(c) and 3(d), we compared the influence of flow velocities (20, 40, 60, and 80 µL/min) on the normalized wavelength shift factor ($\mathrm{\Delta }W/\mathrm{\Delta }{W}_{\max }$), which reflects the relative position of wavelength shift. Here, $\mathrm{\Delta }W$ is defined as the difference between the varying reflection center wavelength and the minimum value, and $\mathrm{\Delta }{W}_{\max }$ is a constant value calculated from the difference between the maximum and the minimum value of the reflection center wavelength. Apparently, the time interval at which the reflective band shifts gradually decreases as the flow velocity increases.

As shown in the insets of Figs. 3(c) and 3(d), the maximum change rate of $\mathrm{\Delta }W/\mathrm{\Delta }{W}_{\max }$, namely the maximum slope of the fitting curves in Figs. 3(c) and 3(d), was expressed as ${\eta }_{\max }$ and also monitored at different flow velocities. E7 with a higher viscosity coefficient would have a larger friction resistance and smaller diffusion coefficient in the porous polymer network, leading to a lower ${\eta }_{\max }$. Conversely, for BA with a lower viscosity coefficient, the diffusion coefficient should be larger, and the shift of the reflection band to the short wavelength was accelerated. The results coincide well with the prediction by analyzing the mutual diffusion distance from Eq. (2). In addition, ${\eta }_{\max }$ was connected with the flow velocity of the injected solution in both cases. The increase of flow velocity also leads to the concentration difference maintained at peak level, which facilitates the diffusion process at the interface between the PCLC film and the fluid. As depicted in Figs. 3(e) and 3(f), the changes of $n_{\mathrm{av}}$ and $p$ dominate the variation of the reflection band^[25]. The swelling process of the network in E7 was possibly slow. Remarkably, in the case of injecting BA, the blue-shift rate of the reflection band was accelerated with the decreasing $p$ in the relatively fast deswelling process.

Furthermore, to exhibit the dynamic coloration by the diffusion of VOCs, the as-prepared UV-PCLC films with retained helical skeleton in double-open-ended glass cells with a thickness of 20 µm were subject to the volatilization of alcohol in a home-made sealed chamber. In the chamber, 10 g of alcohol ($n=1.36$) was put ($10\mathrm{cm}\times 10\mathrm{cm}\times 5\mathrm{cm}$) and volatilized at room temperature, allowing the diffusion of alcohol molecules into the glass cell from one open end and the infiltration of the PCLC film. As shown in Figs. 4(a)–4(d), when alcohol with lower RI was diffused into the film, the wavelength of the main reflection center blue-shifted from $\sim 630\mathrm{nm}$ to $\sim 440\mathrm{nm}$. The variation of reflection spectra can be well explained by Fig. 4(e), which reveals that the diffusion of alcohol in the PCLC film might possibly experience four stages within 10 h.

In Stage a, an unexpected slight red-shift of the main reflection band was observed within the first 30 min, ascribed to the elongation of $p$ due to the swelling of adhesive tapes used to assemble the glass cell. The inset of Fig. 4(a) illustrates the dependence of the thickness of an empty LC cell on the diffusion time of ethanol. The pitch $p$ increased and tended to be stable at a value of $p_1$ after 60 min.

The uneven distribution of the polymer network was caused by UV exposure. The polymer-rich layer could be formed close to the UV light, while the polymer-poor layer is on the other side^[26–28]. In Stage b, as shown in Fig. 4(b), the rise in the two reflection peaks at 455 nm and 590 nm was accompanied by a decrease in the reflection peak at 640 nm, which was caused by the disappearance of the original pitch $p_1$ along with the appearance of two decreased pitches ($p_2$ and $p_3$). Ethanol molecules tended to diffuse along the interfaces between the PCLC film and both substrates. The infiltration of ethanol deswelled the polymer network, initiating the sudden appearance of pitch $p_2$ in the polymer-rich layer and $p_3$ in the polymer-poor layer. As shown in the inset of Fig. 4(b), a similar evolution of reflection peaks was also observed in the polymer-poor film, which was fabricated by UV irradiation for only 2 s, supporting the aforementioned assumption from another aspect. The relationship of three pitches was $p_1>p_2>p_3$ because the pitch contraction depended on the amount of ethanol refilled in different layers.

In Stage c, as shown in Fig. 4(c), the peak at the major reflection band centered at 590 nm was relatively stable at the long wavelength edge due to the slight change of $n_{\mathrm{av}}$ and $\mathrm{\Delta }n$ in the polymer-rich layer. However, the sparse network facilitated the diffusion of ethanol molecules, with a further decrease of $n_{\mathrm{av}}$ in the polymer-poor layer and the blue-shift of the minor reflection center from 445 nm to outside the detection range [Fig. 4(c)]. The diminishment of reflection intensity can be ascribed to the decrease of $\mathrm{\Delta }n$. Finally, although the lateral diffusion in the polymer-rich layer from the open end was not evident, when the concentration of alcohol in the polymer-poor layer reached a suitable value, the occurrence of vertical diffusion into the polymer-rich layer was allowed in the following Stage d. The inset in Fig. 4(d) shows the continuous blue-shift of the central wavelength from 590 nm to 440 nm, reflecting the significant decrease of $n_{\mathrm{av}}$ in the polymer-rich layer.

Additionally, as shown in Fig. 5, the aforementioned approaches were adopted to infiltrate the cholesteric scaffolds in a uniform and a gradient way, respectively, to endow the 100th anniversary logo of Xiamen University (XMU) with vivid structural colors. A microprojection system based on digital micromirror device (DMD) was used to record the XMU logo^[29,30]. In detail, for the approach based on microfluidics, only the logo pattern was fabricated and immobilized in the microchannel. For the VOC diffusion approach in a double-open-ended LC cell, the area outside the logo was cured under UV light at an elevated temperature above the clear point to eliminate structural color, but it still maintained the integrity of the whole network structure in favor of the intact diffusion process of alcohol. By this means, a fascinating iridescence was generated by the volatilization and diffusion of alcohol. To be more specific, the XMU logo appeared with a red color brought about by the increase in the number of layers of $p_2$ and $p_3$ at 35 min (Stage b). Another iridescent logo with gradient colors of orange, yellow, green, and blue was observed at 300 min. More than half of the area showed gradient colors of blue and purple after 600 min (Stage d).

In summary, we disclosed the dynamic coloration of the PCLC network facilitated by two key approaches to refill organic compounds, that is, the alternative injection of mutually soluble organic fluids into a microfluidic channel and the diffusion of organic vapor. For the first approach, the relationship between flow velocity of fluid and the optical response was studied, enabling the reversible tuning of reflection color with a central wavelength located between $\sim 450\mathrm{nm}$ and $\sim 600\mathrm{nm}$. For the second approach, the influence of the duration time in UV polymerization on optical characteristics was studied and interpreted. A VOC-diffusion-induced blue-shift of the reflection center was revealed from $\sim 620\mathrm{nm}$ to $\sim 410\mathrm{nm}$. We anticipate that the class of polymer-templated LC composites with dynamic coloration by infiltrating the cholesteric scaffold with organic compounds can become one of the most promising candidates for dynamically tunable optofluidic reflectors, visually readable sensors, or compact anti-counterfeit labels in response to VOCs.