Switchable biomimetic nanochannels for on-demand SO2 detection by light-controlled photochromism

In contrast to the conventional passive reaction to analytes, here, we create a proof-of-concept nanochannel system capable of on-demand recognition of the target to achieve an unbiased response. Inspired by light-activatable biological channelrhodopsin-2, photochromic spiropyran/anodic aluminium oxide nanochannel sensors are constructed to realize a light-controlled inert/active-switchable response to SO2 by ionic transport behaviour. We find that light can finely regulate the reactivity of the nanochannels for the on-demand detection of SO2. Pristine spiropyran/anodic aluminium oxide nanochannels are not reactive to SO2. After ultraviolet irradiation of the nanochannels, spiropyran isomerizes to merocyanine with a carbon‒carbon double bond nucleophilic site, which can react with SO2 to generate a new hydrophilic adduct. Benefiting from increasing asymmetric wettability, the proposed device exhibits a robust photoactivated detection performance in SO2 detection in the range from 10 nM to 1 mM achieved by monitoring the rectified current.

The modification of SP probes to the aminated surface of AAO nanochannels using the diffusion-limited patterning method (DLP). The ethanol solution of SP probes was arranged on the porous side, while pure ethanol solution was arranged on the opposite barrier layer side.
The diffusion-limited patterning (DLP) model is a unique technique for patterning functionalized molecules on surfaces of the nanofluidic channels. [S3] The modification was realized by the diffusion of the functional molecules in a solution phase. Herein, an APTES-treated AAO membrane with tubular nanochannel geometry was mounted between the two chambers of the H-shaped electrochemical tank ( Supplementary Fig.  6). The SP ethanol solution (containing EDC and NHS) was added to the porous side of AAO, while pure ethanol solution was added to the opposite barrier layer side. In this case, a concentration gradient was built along the nanochannel axis from the porous side to the barrier layer. As a result, SP molecules could spread from a high concentration (porous side) to a low concentration (barrier layer side). During the process of diffusion, the SP molecules could be chemically conjugated with the inner surfaces of the nanochannels, leading to the successful patterning of functionalized SP to the surface of AAO nanochannels. Herein, SP molecules were successfully immobilized onto a porous segment of the tube-shaped AAO nanochannels, but not on the blocking layer side. This results in the formation of asymmetric SP/AAO nanochannels. Importantly, by adjusting the time of patterning, SP/AAO nanochannels' surface properties including surface charge and surface wettability could be exactly controlled. Fig. 7 SEM images of the top and bottom surfaces of the AAO nanochannels modified by (a, b) APTES and (c, d) SP molecules. In the APTESmodified AAO and SP/AAO nanochannels, we could observe a porous structure on the top surface and a dense hemispherical dome structure on the bottom surface, which was similar to the pristine AAO. It was concluded that the AAO nanochannels were not damaged during the attachment process. Supplementary Fig. 8 Detailed deconvoluted X-ray photoelectron spectra of (a) N1s and (b) C1s for the SP/AAO nanochannels. X-ray photoelectron spectra (XPS) is a versatile surface analysis technique for determining the elemental composition information of the materials. Here, detailed deconvolution XPS spectra of N1s and C1s were applied to identify the successful chemical immobilization of SP on the APTES-treated AAO nanochannels. [S4] The curve fitting was performed by PHI MultiPak software, and Gaussian-Lorentz functions and Shirley background were used. As shown in Supplementary Fig. 8a, the deconvoluted N1s spectrum showed two characteristic peaks at 399.9 eV and 405.8 eV, which are assigned to C-N/O=C-N bonds from the indoline/amide nitrogen and N-O bonds from the nitro group, respectively, indicative of the presence of SP components as expected. However, no characteristic peak of the ring-opening MC indoline fraction (N + component) was observed at 400.5-401.0 eV. [S5,S6] These results suggested that the SP probes on the SP/AAO nanochannels had not translated into MC form during XPS measurements. In the C1s core level region of Supplementary Fig. 8b, the XPS C1s spectra were fitted into three peaks at 284.8 eV, 286.4 eV, and 288.4 eV, which are associated with C-C/C-H, C-N/C-O, and O=C-N groups, respectively. The high-resolution C1s peak from SP/AAO nanochannels confirmed the chemical grafting of SP probes, consistent with the above N1s result. It is worth noting both the nitrogen peak found at approximately 399.9 eV and the carbon peak found at approximately 288.4 eV for the SP/AAO membrane belonged to the amide groups. [S7-S9] This indicated the covalent functionalization of SP probes on APTES-modified AAO surface via EDC-NHS chemistry. All of these results verified that SP probes were successfully immobilized onto the channels by covalent binding.

Supplementary
Supplementary Fig. 9 Zeta-potential changes of the AAO nanochannels at different modification times of SP probes.
Zeta potential is a model parameter describing the surface charged behaviour including electrical properties and the number of charges at the solid-liquid interface. As the data in Supplementary Fig. 9 demonstrated, the APTES treated-AAO nanochannels exhibited the highest positive values in zeta potential. The results indicated that abundant positive charges on the nanochannel surface appeared, which should be attributed to the protonation of the terminal amino groups from APTES in the slightly acid environment. Then the zeta potential of the SP/AAO nanochannels decreased with the increase of SP modification time from 0 to 3 h. This trend indicated that neutral SP components were gradually deposited on APTES-treated AAO as the modification time extended. As a result, the amino positive charge of APTES and the hydroxyl positive charge of AAO would be gradually covered, resulting in a decreasing positive charge density. Benefiting from the confinement effect in nanoscale channels, the reduction of the positive charge density would degrade the ion selectivity and ion rectification of the nanosystem. The overall zeta potential trend well verified the surface charge variation and ion transport behaviour across the nanochannel membrane. Supplementary Fig. 10 CA of the bottom surface of APTES-treated AAO nanochannels at different modification times of SP probes. The CA of the barrier layer on the bottom surface is less changed in the process of patterning SP to the nanochannel surface using the DLP method. This is because the SP solution on the porous side cannot diffuse to the blocking layer side of AAO due to the closed morphology of the blocking layer. Supplementary Fig. 11 Optical photograph of the SP/AAO nanochannels after 10 minutes of exposure to (a) blue (450 nm, 20 mW), (b) yellow (590 nm, 6.4 mW), (c) red (660 nm, 12.6 mW), and (d) infrared light (808 nm, 14.5 mW). After 10 minutes of visible and infrared light irradiation, no colour change was observed in the SP/AAO nanochannels, suggesting that UV irradiation did not convert SP molecules to the MC form. Supplementary Fig. 12 The absorption changes of SP/AAO nanochannels during UV irradiation and SO2 exposure.
Before irradiation, the spiropyrane molecules were predominantly in the closed-ring SP form. As shown in Supplementary Fig. 12, the SP-state nanochannels did not have any absorption band in the visible region. After the UV stimulated conversion to the MC purple isomer, the MC-state system incorporated a charge-separated zwitterionic conjugated structure. At this point, we observed a high absorption peak at 550 nm, which is a characteristic peak of the MC. [S10] Subsequently, when in contact with SO2, the peak at 550 nm fully disappeared. This was because the formation of the new MC−SO3H species destroyed the conjugated structure and electron configuration of MC. Such absorption phenomena are consistent with changes in contact angle, fluorescence, and optical colour during UV irradiation and SO2 exposure. Together, the results demonstrated that UV stimulus could control the photoconversion of SP to MC enabling SO2 binding into the nanochannel. Therefore, the as-prepared SP/AAO nanochannels could be exploited for real-time signalling of SO2 for on-demand sensing applications in a light-controlled manner. Supplementary Fig. 13 S2p peak of X-ray photoelectron spectra for the UV-activated SP/AAO nanochannels before and after interaction with SO2.
As demonstrated in Supplementary Fig. 13, before and after the SO2 response, we observed a peak on the SP/AAO nanochannels with a binding energy of about 168.4 eV, which is the characteristic peak for S2p elements. [S11] It is important to explain that the sulfur element that appears before the SO2 addition originated from the sulfuric acid electrolyte used in the preparation of the AAO nanochannels. However, after being treated with SO2, a significant increase in the sulfur peak was observed. As presented by XPS atomic ratio, the S atomic percentage in the SP/AAO nanochannel increased from 1.02 to 1.83 after the reaction with SO2 (Supplementary Table 2). The result is further proof of the successful nucleophilic attack of SO2 on MC components of the functionalized nanochannels. Supplementary Fig. 14 Reaction mechanism between SO2 and C=C of the MC form in the nanochannels. Supplementary Fig. 15 CA of the bottom surface of SP/AAO nanochannels during UV-controlled response to SO2. After UV stimulus and SO2 detection, no changes in CA at the bottom blocking layer were observed due to the lack of SP components and response behaviour on the bottom surface. Supplementary Fig. 16 The MS analysis of the reaction between SP and SO2 in solution (a) in the absence of UV and (b) in the presence of UV.
To further demonstrate that the light-controlled inert/active switching function can be used for on-demand detection of SO2, the response of the SP probe and SO2 with and without UV irradiation was characterized by using mass spectrometry in solution. As shown in Supplementary Fig. 16a, the resulting mass spectrum in the absence of UV irradiation exhibited a peak at m/z = 381.7440, which was attributed to the SP. The results clearly showed that no new product was formed between SP and SO2, i.e., the original SP was chemically inert to SO2. However, when the SP solution was subjected to UV irradiation prior to the SO2 treatment, we observed a new peak at m/z = 461.0899 indicative of [MC−SO3H -H + ] (Supplementary Fig. 16b). The generation of the MC−SO3H compound after UV illumination confirmed the addition reaction between SO2 and the photoisomer MC of SP. Accordingly, the comparative results of (a) and (b) provided evidence for the successful light-controlled reaction behaviour of SP towards SO2. It was concluded that the specific reactivity of MC to SO2, but not SP to SO2. Supplementary Fig. 17 UV response current (IUV) of SP/AAO nanochannels at different times. Error bars denote the standard deviation from three different samples. Supplementary Fig. 18 Optical photograph of SP/AAO nanochannel membrane at incremental UV irradiation times. As the UV radiation time increases from 0 to 10 min, the colour of the membrane gradually changed from colourless to pink to purple. However, the colour of the nanochannels stopped changing after 10 min. This intuitionistically reflected the photochromic shift of the SP/AAO nanochannels and reached stability within 10 min. The result was consistent with the trend of the UV response current. Fig. 19 Effect of temperature on (a, b) MC-state SP solution and (c, d) MC-state SP/AAO nanochannel membrane.

Supplementary
It has been reported that heating and/or visible light irradiation can convert MC back to SP. [S12] Firstly, we tested the effect of temperature on liquid MC in the darkroom. The optical photographs clearly demonstrated the gradual fading of the colour of MC solution with increasing temperature from 25, 37, 45 to 60°C (Supplementary Fig. 19a).
This phenomenon verified that the coloured MC reverted to colourless SP under heating conditions. In addition, the fluorescence data further proved this effect. As shown in Supplementary Fig. 19b, the fluorescence intensity decreased as the temperature increased, indicating that heating led to a gradual reduction of the MC component in the solution.
However, once the molecules were immobilized on the AAO substrate, the MC components were relatively stable under heating conditions. As in Supplementary Fig.   19c, there is no significant colour change on the surface of the MC-state SP/AAO nanochannels at different temperatures of 25, 37, and 45 °C. The approximate effect of heating on the ion transport behaviour implied that the surface composition of the channels is in a similar state in all three cases ( Supplementary Fig. 19d). As the temperature increased to 60°C, we observed a slight discoloration of the membrane. At this point, a small difference in ionic current across the membrane was recorded. This should be due to the return of a tiny amount of MC on the nanochannels to the less hydrophilic SP state. These phenomena indicated that temperature made little influence on the solid SP/AAO nanochannel sensor. The overall comparative results showed that our SP/AAO nanochannels can be applied in a wide range of SO2 detection than liquid probes. Supplementary Fig. 20 The modification of AP probes on the inner surface of AAO nanochannels to build a nonswitchable AP/AAO nanochannel sensor.
First, 3-isocyanopropyltriethoxysilane (IPTS, another silane coupling agent with bilateral functional groups) was modified to the inner surface of the AAO channel. Its terminal isocyanic acid was able to react with the amino group. Thus, the AP molecules with the amino groups could be chemically immobilized on the IPTS surface using the DLP technique. The diffusion time was set to 10 h to achieve uniform modification. Supplementary Fig. 21 The response performance of the AP/AAO nanochannels to SO2 solution.
In 2020, Sun et al. developed a high-efficiency bisulfite nanosensor by immobilizing aminobenzophenone (AP) on the wall of a single conical nanochannel in a polyethylene terephthalate (PET) membrane. [S13] After completion of immobilization, the appearance of blue fluorescence on the cross-section of the nanochannel proved the successful modification of the AP probes. The ketone group of the probes on the biosensor surface could specifically interact with HSO3 − via a nucleophilic reaction, followed by a significant increase in the hydrophilicity and ion transportation. Simulating the intelligent nanofluidic biosensors described above, we constructed here a porous AP/AAO nanochannel system. As shown in Supplementary Fig. 21, a blue fluorescence signal appears in the AP/AAO nanochannels, which was direct evidence of the successful modification of AP probes. The SO2 response result also described a similar trend. After HSO3solution treatment, we observed a significant increase in the ionic current and surface wettability of AP/AAO nanochannels ( Supplementary Fig.  21a, b). In addition, the response behaviour of AP and SO2 in the nanochannel spaces was tracked by LSCM. As shown in Supplementary Fig. 21c, the fluorescence signal decreased dramatically after the reaction. This indicated that SO2 could efficiently attack the ketone group on the probe, leading to the disruption of the conjugated structure of the AP. In conclusion, these combined results demonstrated the successful construction of an AP/AAO nanochannel sensor for SO2 detection.