Photoswitchable Binary Nanopore Conductance and Selective Electronic Detection of Single Biomolecules Under Wavelength and Voltage Polarity Control

Abstract

We fabricated photo-regulated thin-film nanopores by covalently linking azobenzene photoswitches to silicon nitride pores with ~10 nm diameters. The photo-responsive coatings could be repeatedly optically switched with deterministic ~6 nm changes to the effective nanopore diameter and of ~3× to the nanopore ionic conductance. The sensitivity to anionic DNA and a neutral complex carbohydrate biopolymer (maltodextrin) could be photo-switched “on” and “off” with analyte selectivity set by applied voltage polarity. Photo-control of nanopore state and mass transport characteristics is important for their use as ionic circuit elements (e.g. resistors and binary bits) and as chemically tuned filters. It expands single-molecule sensing capabilities in personalized medicine, genomics, glycomics, and—augmented by voltage polarity selectivity—especially in multiplexed biopolymer information storage schemes. We demonstrate repeatably photo-controlled stable nanopore size, polarity, conductance, and sensing selectivity—by illumination wavelength and voltage polarity—with broad utility including single-molecule sensing of biologically and technologically important polymers.

Solid-state abiotic nanopores capture the imagination and have emerged in applications as high-performance platforms for single-molecule science, as nanoscale apertures for fundamental physics experiments and controlled cargo delivery, as conductive and often rectifying ionic circuit elements, as high resolution model systems for nanoporous filters, and as robust, device-ready analogues and mimics of their proteinaceous brethren.^1-6 In most of these undertakings a nanopore—a ~10 nm-diameter channel through a ~10 nm-thick membrane—is immersed in an ionic electrolyte and provides the only path for voltage-driven mass transport between solutions on either side of the membrane. In the most prevalent application of single-molecule DNA sequencing, electrophoresis of a DNA strand through a nanopore alters the ionic current flow to give rise to characteristic signals that can be used to recover the DNA base sequence.⁷

Photo-responsive nanopores are even more appealing than conventional pores. Nonlocal, noncontact, and wavelength-selectable control of photo-responsive nanopores can significantly enhance their properties and capabilities. Molecular photoswitches that respond to light by configurational isomerization are a prominent tool that can, when suitably coupled to a nanopore, be used to optically control aspects of its nanoscale structure and function. This approach is in contrast to those using conventional nanopores to give photoswitch-configuration-specific readout of target molecules or complexes functionalized with photochromophores.^8-11 Chemically derivatizing the nanopore surface, rather than the molecule, allows nanopore sensing to remain truly label-free, places the responsibility for chemical functionalization in the hands of the device fabricator rather than the practitioner, provides a fixed route to surface functionalization (vs. the myriad of approaches that might be required to derivatize different molecules), and decreases the time needed for sample preparation for nanopore sensing.⁵ At the same time, chemical functionalization of the nanopore rather than the molecule is better suited to applications, such as filtering or controlled delivery, that require the platform, itself, to deliver the desired structure and performance. Photoswitches including spiropyrans,^{12, 13} hydrazones,¹⁴ donor-acceptor Stenhouse adducts,¹⁵ stilbenes,¹⁶ and azobenzenes^{17, 18} isomerize upon light irradiation, which leads to changes to their structure, polarity, and various electronic and photophysical properties. Such molecular-scale changes to the nanopore coating can have longer-range effects including altering interfacial properties (e.g. the Debye-layer) that in turn control dynamic processes like nanoscale mass transport through the nanopore.^{3, 5, 19-22} They can thus modulate the sensing of the ion, small- and macromolecule, and nanoparticle and virus targets of nanopore sensing.

We chose both nanopore platform (silicon nitride) and photo-switch (azobenzene) to be robust and adaptable for a wide range of downstream applications. Low-pressure chemical vapor deposition (LPCVD) silicon-rich silicon nitride (${\text{SiN}}_{\mathrm{x}}$) is the most prevalent and conventionally versatile nanopore fabrication material supporting nanopores of a wide range of sizes.^{1, 2, 5, 20, 23, 24} It is a standard material in microelectronics so that the integration of nanopores into complex devices can be conceived of alongside manufacturing at-scale. Azobenzenes, in particular, display facile reversible switching with alternating UV and visible light irradiation between the ground-state trans isomeric form and its metastable cis isomer counterpart (Figure S1). The cis-to-trans (reverse) isomerization can be also promoted by thermal activation, offering an additional method of switching. Due to the ease of synthesis and derivatization,²⁵ high cyclability of the photoswitching processes,¹⁸ high isomerization rates,¹⁷ and significant structural and polarity changes,^{26, 27} the azobenzene derivatives are widely incorporated in various material systems, notably polymeric films,^28-30 nanoparticles,^31-33 and cargo-containing vesicles.³⁴ Configurational isomerization inside a zeptoliter volume such as a nanopore or in nanocages, nanoporous frameworks, or nano-cavities between aggregated nanoparticles³⁵—where even the solvent structure and energetics may be perturbed by confinement—is itself a rich area for investigation. We anticipated, however, that a suitable azobenzene covalently coupled to the surface of a thin-film nanopore (Scheme 1) and remaining photo-switchable would be able to effectively adjust the size and polarity of nanopore systems. We first assessed whether we could reliably photoselect each azobenene configurational isomer and what the resultant nanopore properties would be. We then challenged azobenzene-festooned nanopores with two analytes—DNA and the complex carbohydrate maltodextrin. Each presented distinct physicochemical properties to the nanopore and each was drawn from a molecular class core to pursuits within genomics and glycomics, respectively.^{5, 19, 36}

Scheme 1. Formation and coating of ${\mathbf{SiN}}_{\mathbf{x}}$ nanopores.

Nanopores were fabricated by voltage-induced controlled (dielectric) breakdown (CBD) in buffered electrolyte and then covalently functionalized with the azobenzene photoswitch by photohydrosilylation in acetonitrile using 254 nm light. Longer wavelengths of light present in unfiltered mercury lamps can drive trans→cis isomerization.

RESULTS AND DISCUSSION

The 4-(propargyloxy)azobenzene reversibly photo-switched for more than 100 cycles in free solution (Figure S2). The half-life of the metastable cis 4-(propargyloxy)azobenzene in free solution at room temperature was 61 h in the organic solvent (acetonitrile) used for photo-switching and 41 h in the aqueous electrolyte (1 M KCl/10 mM HEPES) used for characterization measurements (Figure S3). We therefore expected no (bulk) solvent-induced impediments to photochemically and thermally stable isomerization of our nanopore coating. The nanopore itself, however, may present barriers to photo-switching from confinement, surface curvature, or interactions with the inorganic surface. Unlike a photochromic coating on the outer surface of a nanoparticle,^{37, 38} for example, where the free end of the molecule is further away from its neighbors, the attachment of a molecule inside a nanopore means that the free ends of the molecule will all tend to be more sterically crowded than at the surface. Interactions arising from crowding in dense films may also affect the nature of the photoswitching.

A ${\text{SiN}}_{\mathrm{x}}$ nanopore of (original) diameter $d_{{\text{SiN}}_{\mathrm{x}}}$ coated with a molecular film of thickness $t_{\text{film}}^{cis∕trans}$ will have a physical diameter of $d_{\text{film}}^{cis∕trans}=d_{{\text{SiN}}_{\mathrm{x}}}-2\cdot t_{\text{film}}^{cis∕trans}$ with $t_{\text{film}}^{trans}$ up to 1.47 nm and $t_{\text{film}}^{cis}$ up to 1.05 nm (Figure S1, 1). Nanopore dimensions dominate the nanopore conductance, $G$, in molar ionic strength solutions

$$G=\sigma {\left(\frac{4l}{\pi d^2}+\frac{1}{d}\right)}^{-1}$$ (1)

where $d$ and $l$ are the nanopore diameter and length, respectively, and $\sigma$ is the solution conductivity.^{23, 39} The nanopore conductance is then dependent on the cis- vs. trans-dependent thickness of the surface coating. All current (and thus conductance) measurements were performed in the dark. The measured conductances before and after the installation of a film or after a photochromic reaction can then be used to infer the film thickness $t_{\text{film}}^{cis∕trans}=\frac{1}{2}(d_{{\text{SiN}}_{\mathrm{x}}}-d_{\text{film}}^{cis∕trans})$ or change in film thickness $(\mid \Delta d_{film}\mid =2\mid t_{\text{film}}^{cis}-t_{\text{film}}^{trans}\mid )$, respectively. The functionalized nanopores showed reversible photo-switching between a high and a low conductance state (Figure 1, Figure S4) as would be expected from cycling of the film thickness between $t_{\text{film}}^{trans}$ and $t_{\text{film}}^{cis}$ (Figure 1) and the conductance between $G_{trans}$ and $G_{cis}$. The high conductance state was induced by 365 nm light—the wavelength that drives the trans→cis isomerization that would increase the pore diameter and conductance (by Equation 1). The low conductance state was induced by visible light—this excitation drives the cis→trans isomerization reaction that would decrease the pore diameter and conductance. The conductance data and wavelength-dependence of switching show that the pore coating is initially predominantly in the cis configuration despite using the more energetically stable trans isomer for photohydrosilylation. Separate bulk solution-based experiments show that the spectrum of the unfiltered mercury lamp used for photohydrosilylation could also drive trans→cis isomerization (Figure S5), albeit at a low ~20% trans→cis conversion yield in bulk solution (~55 mM). The apparent enrichment of the cis content of the surface coating over the bulk could be explained by the more hydrophilic nature of the cis isomer (vide infra) causing preferential transport to, or partitioning into, the nanopore interior when compared to the trans isomer. It could also be explained by the more prosaic mechanism that an initial trans film is converted to the cis form over the course of the photoirradiation. With the surface coating in place the reversible cycling of photo-induced changes in conductance is in accord with the photochemical reaction mechanism in Figure 1 and convincingly supports the successful formation of a photochromic ${\text{SiN}}_{\mathrm{x}}$ nanopore.

Figure 1.

(a) Photoswitching of the azobenzene film coating by 365 nm (UV flashlight, trans→cis) and white light source (cis→trans) should drive physical changes in monolayer effective thickness and thus nanopore diameter (dashed circle). Photoirradiation was ceased prior to measuring currents. (b,c) The initial conductance of the nanopore after coating (“azobenzene”) was recorded before the photoswitching experiments with the conductance measured after each photoirradiation step. The measurements revealed two statistically significant conductance states.

The average conductance-derived diameter change (Equation 1) upon installation of the 4-(propargyloxy)azobenzene was $\overline{d}\pm \sigma =1_{\cdot 8}\pm 1_{\cdot 3}\text{nm}$ (8 independent pores), corresponding to a ~1 nm film thickness. This is consistent, within typical measurement accuracy^{20, 40, 41} with the installation of no more than a monolayer of primarily cis-4-(propargyloxy)azobenzene ($t_{\text{film}}^{cis}=1.05\text{nm}$, Figures 1, S1). Wetting effects, surface coverage variability (vide infra), and differences in relative cis/trans enrichment could account for the range of initial apparent film thicknesses. Comparison of Figures S4 and S6 shows that despite this initial variability all of the functional coatings could be definitively and controllably photo-switched between two readily distinguishable high and low states. Switching between the 1.47 nm trans film thickness and the 1.05 nm cis film thickness (Figures 1, S1) would change the physical nanopore diameter by 0.84 nm and this would change the conductance in accordance with Equation (1). The mean conductance-derived diameter change, derived with Equation (1) and the conductance changes across 8 independent pores was $\Delta d_{\text{switch}}=6_{\cdot 3}\pm 1_{\cdot 1}\text{nm}$. This corresponds to a ~3 nm change in apparent film thickness accompanying the cis/trans isomerization, approximately 7× the magnitude expected from the change in molecular dimensions. In earlier nanopore surface coating work,²⁰ 4× greater increases than the molecular dimension in conductance-derived film thickness were ascribed to increases in the hydrophobicity of the surface coating. We present density functional theory (DFT) calculations below that show that the trans isomer is indeed more hydrophobic than the cis isomer, so that the numerical discrepancy does not compromise the validity of the connection of conductance switching to molecular isomerization but rather underscores the importance of wetting in the often challenging nature of nanoconfined chemistry.^{42, 43} When nanopore coatings are hydrophobic, the diameter calculated from the conductance using Equation (1) is better treated as an effective diameter—incorporating size and wetting—than a physical diameter. Importantly, this complication highlights that this solid-state photochromic nanopore may be useful as a platform not just for applications but for exploration of such fundamental phenomena.

The 4-(propargyloxy)azobenzene moiety offers two distinct nanopore coatings without change of chemical formula or molecular structure other than isomerization about a double bond. Unlike for some photochromic species such as spiropyrans, there is no change in the discrete formal charges on the molecule during isomerization in the pH range of interest (${\text{pK}}_{\mathrm{a}}\mathrm{H}$ trans-azobenzene −2.95,⁴⁴ cis-azobenzene −1.49⁴⁵). Each configuration of the azobenzene, however, presents a different charge distribution (Figure S7) at the interface between the nanopore surface and solution—and to any proximal analyte molecule. The consequences of isomerization thus transcend geometry. The trans isomer is largely hydrophobic with a dipole moment of 2.1 D whereas the cis isomer is more hydrophilic with a dipole moment of 3.9 D, according to DFT calculations of each isomeric structure. This large polarity change of the azobenzene coating will affect the nanopore wetting, in accordance with previous reports where the photo-isomerization of surface-functionalized azobenzene changed the contact angle of liquid on the surface.^46-48 In addition, 4-(Propargyloxy)azobenzene in the lower energy state (trans) has limited solubility in aqueous solutions, as confirmed by the strongly reduced solution absorbance in the UV-vis spectra (Figure S8a). The solid undergoes UV-induced isomerization and then, as a result of the increased hydrophilicity of the cis isomers, undergoes subsequent dissolution in the aqueous solution. The absorption spectrum of the cis species in aqueous solution displays the characteristic absorption bands of a cis state (π–π* at 310 nm and n–π* at 430 nm) consistent with those recorded in organic solutions (Figure S8b). The solubilized cis isomers switch back to trans upon white light irradiation and immediately precipitate (Figure S8a). Thermal activation at 75 °C also leads to the facile reversion to trans which precipitates upon cooling to room temperature (Figure S8c). These observations in bulk solution are useful to provide insight into how the properties of each isomer could affect the interface between the surface coating and the confined ionic solution. Differences in nanopore wetting can affect the apparent nanopore size in ways not explicitly accounted for in Equation (1) where the more hydrophobic trans-azobenzene-coated nanopore appeared smaller than would be explained by the molecular dimensions, alone (vide supra). The effect of coating-induced restructuring of the nanopore-entrained electrolyte could be accounted for by an effective solution conductivity in Equation (1). The hydrophilicity difference between the two isomers also offers a physicochemical basis for the apparent cis enrichment of the nanopore coating upon photohydrosilylation (vide supra). In spite of the potential and observed solvation complexities we could reversibly switch the photochromic pore in pure water (Figure S9) and without (conductance-inferred) differences in photoisomerization reaction endpoints versus acetonitrile.

An unfunctionalized ${\text{SiN}}_{\mathrm{x}}$ nanopore surface is comprised of a mixture of surface —OH and —NH₂ groups that make it neutral at pH ~4.3 (by —O⁻ and —NH₃⁺), positively charged at more acidic pH (by —OH and —NH₃⁺) and negatively charged at larger pH values (by —O⁻ and —NH₂).^20,49 This change in surface charge with pH changes the conductance at the same time.²⁰ For ${\text{SiN}}_{\mathrm{x}}$, there would be a minimum in the conductance at the isoelectric point (pI ~4.3) with the higher conductance values at more and less acidic pH values when the surface is charged. We measured the conductance of the same photochromic nanopore in both configurations as a function of pH (Figure S10a,b). The transition between two regions of opposite slope, coupled with measurements using maltodextrin at pH 7 (vide infra)¹⁹, indicate a positively charged surface in the less acidic region for the cis pore. The measured curves show no dramatic departure from the general behavior of the bare ${\text{SiN}}_{\mathrm{x}}$ pore surface. In addition, cis 4-(propargyloxy)azobenzene was observed to be stable at pH 3 over 110 minutes (Figure S10c) supporting the robustness of the cis pores and studies performed at low pH conditions. Since the 4-(propargyloxy)azobenzene is neither acidic nor basic and does not form charged species within pH 3-7,⁵⁰ it is more likely that the surface coating density is insufficient to prevent the underlying chargeable ${\text{SiN}}_{\mathrm{x}}$ from being exposed to solution and then contributing to the pH-dependent conductance. This is in contrast to earlier work with short (compact) alkyl chains where it was the surface coating properties that dictated the change of conductance with pH, with no indication of exposed ${\text{SiN}}_{\mathrm{x}}$.²⁰ The trans film will create a more hydrophobic local surface environment for any unprotected ${\text{SiN}}_{\mathrm{x}}$ and the smaller pore diameter will amplify its influence on the nanoconfined solution structure which is the means by which the surface charge is detected through conductance. This local hydrophobic environment could explain why the experimental cis and trans curves are shifted relative to each other and the curve for the trans film showed less concordance with the curve for bare ${\text{SiN}}_{\mathrm{x}}$ than the cis film.

We created nanopores with photoselectable size, polarity, and conductance and, importantly for applications, reliable and stable operation. Photo-cycling between the two distinct low (trans) and high (cis) conductance states of the pore was done in discrete experimental steps—photo-switching in acetonitrile buttressed by conductance measurements in aqueous electrolyte before and after—that demonstrated both the reversibility and the robustness of the photochromic nanopores to solvent exchange and mechanical handling. Redox-mediated electrocatalytic Z-to-E reversion processes require higher voltage conditions, at least 1 V versus Fc/Fc⁺, than the 200 mV used in this work.⁵¹ Significantly, the two distinct conductance states, supporting up to ~20 nA current flows through the pore, were stable across the 30 min measurement time and in the presence of high electric fields (up to ~200 mV/15 nm≅13 mV/nm) across the nanopore. Isomerization switching persisted over 2 weeks before declining in reliability. It is thus clear that we had successfully created a photochromic thin-film nanopore that maintained the metastable states, trans and cis, during and across several measurement periods. Such a nanopore can be used as a digital bit for reversible nondestructive data storage with readout of the state by ionic conductance and setting of the state by photons of the appropriate wavelength.⁵² A photochromic nanopore may also offer technical advantages as the photo-switchable readout element for DNA (and other biopolymer) data storage schemes.^{1, 53, 54} For example it may offer the ability to turn readout off until switched on, or to read the same biopolymer through two different states of the same pore for added signal discrimination. We are also interested in the potential biomedical applications of the photochromic nanopore, in particular for biopolymer characterization and sequencing in the domains of genomics and glycomics.^{5, 7, 24, 36} The complex carbohydrates can take on a greater range of chemical composition, charge, and structure than DNA and this overall biopolymer diversity can be exploited at this exploratory stage to help characterize the photochromic nanopore, itself.¹⁹

For biopolymer analysis by conventional resistive-pulse nanopore sensing the nanopore must deliver fixed size and stable baseline current in the absence of analyte, and the analyte must be able to perturb the open-pore current. Both cis and trans pores gave steady baseline currents suitable for single-molecule sensing and consistent with a structurally stable surface coating (Figures 2, S11, and S12). Noise characteristics were similar to those from unfunctionalized silicon nitride nanopores (Figure S13). Single-molecule-characteristic discrete blockages could be detected for both DNA (+200 mV) and maltodextrin (−200 mV) with the pore in the cis configuration (Figures 2, S12) but the trans pore was nonresponsive to both, regardless of applied voltage polarity (±200 mV; Figures 2, S11). The distinct current blockage transients for the cis pore—downward spikes for DNA and upward spikes for maltodextrin in Figure 2—were isolated by thresholding against the temporally local average current and representative examples are shown in Figures 2 and S14.⁵⁵ A more comprehensive representation of the 418 events (DNA) and 761 events (maltodextrin) is given in Figure 3. The average shifted histogram (ASH) plot of the change in nanopore conductance ($\Delta G$) induced by passage of DNA through the pore, for example, shows multiple peaks that are consistent with passage of non-linearized DNA. The overall signatures of these events are typical of nanopore sensing of DNA and maltodextrin using unfunctionalized silicon nitride nanopores (Figure S15, S16) so that we can conclude that these photochromic nanopores are useful for profiling these two biopolymers with the additional capability of photo-controllable responsivity.^{19, 56}

Figure 2.

The cis and trans configuration of the nanopore coating turned the nanopore sensing on and off for two different molecules: maltodextrin, a neutral complex carbohydrate polymer, and 3 kbp dsDNA, an anionic biopolymer. The nanopores were isolated from the corresponding light sources after photoswitching. Current traces were acquired at +200 mV for DNA with a 10_.1 nm (7_.2 nm) diameter cis (trans) configuration pore and at −200 mV for maltodextrin with a 14.9 nm (7_.6 nm) diameter cis (trans) configuration pore. Absence of analyte gave a steady baseline current. Addition of analyte to the pores with monolayer in the cis configuration resulted in readily detectable current spikes characteristic of that analyte. No such single molecule detection events were apparent with the coating in the trans configuration. Several representative events are shown as an inset.

Figure 3.

We sensed 3 kbp dsDNA using a 10_.1 nm diameter cis nanopore at 200 mV and maltodextrin using a 14.9 nm cis nanopore at −200 mV. All measurements were in 1 M KCl buffered to pH 7 with 10 mM HEPES. Top row. The average shifted histogram (ASH) of the change in the conductance shows peaks corresponding to the open-pore current at $\Delta G=0\text{nS}$ and analyte-induced blockages (peaked at $\Delta G>0$). Discrete events were first extracted by thresholding from the baseline and $\Delta G$ is measured for each event relative to its local baseline. The shaded curves are fits to the experimental data. Middle row. Scatter plots of all isolated events are superimposed with heat maps showing the frequency of events in each analyte. The abscissa, $\langle i\rangle ∕\langle i_0\rangle$, is the mean blockage of each event scaled by the mean open pore current proximal to the event. Bottom row. ASH plots of event duration for all isolated events peak at ~150 and 60 μs for DNA and maltodextrin, respectively.

The voltage polarity to detect DNA in these experiments is consistent with net transport in the direction of electrophoretic motion, as is generally the case for nanopore DNA sensing.⁵⁶ A reduced and unsteady baseline current is common in DNA sensing using unfunctionalized ${\text{SiN}}_{\mathrm{x}}$ nanopores.^{19, 57} It is generally ascribed to “sticking” of the DNA to the nanopore surface and a plethora of surface chemical approaches have been developed to prevent such sticking.^{19, 21, 41} While the 4-(propargyloxy)azobenzene coating does not prevent occasional sticking, further modification of azobenzene moiety with various functional groups, e.g. fluorous groups, could chemically tune this behavior without sacrificing the photo-switchability.^58-60 In an electrokinetic sense, maltodextrin is the antithesis of DNA: it is neutral and therefore nonresponsive to electrophoresis. Instead, transport—and thus sensing—of maltodextrin occurs via electroosmosis that requires a charged nanopore surface.^{5, 19, 56} The voltage polarity for sensing maltodextrin—opposite to that of electrophoresis for anionic DNA—indicates a negative surface charge. An incompletely covered ${\text{SiN}}_{\mathrm{x}}$ surface, such as that suggested by the data in Figure S10, would be terminated with —O⁻ and —NH₂ and thus be negative at the sensing pH of 7.^{20, 49} The consequence of nanopore photo-switching and analyte-specific electrokinetic transport mechanism means that to successfully detect the biopolymer of interest required photo-switching the nanopore to the correct cis configuration and setting the voltage to the analyte-specific voltage polarity.

The photochromic nanopore is nonresponsive to DNA at the voltage polarity where it is responsive to maltodextrin (−200 mV) because the electroosmotic driving force is insufficient to overcome the opposing electrophoretic driving force. Electroosmotic flow can be moderated by electrolyte salt composition and by surface chemical tuning^{19-21, 49, 56}—by pH as suggested by Figure S10 or by surface coating derivatization introduced above in the context of creating non-stick surfaces. There are thus additional, chemical parameters for tuning nanopore responsivity and selectivity in addition to the wavelength and voltage polarity. The broad similarity of conductance vs. pH curves for cis and trans configurations suggests that it is not a loss of nanopore surface charge at pH 7 that makes the trans nanopore nonresponsive to the neutral maltodextrin. Rather, the loss of response to both analytes suggests an origin in the physical size of the pore and hydrophobicity differences. Owing to the hydrophobic nature of the trans coating, the trans nanopores are physically smaller than in their corresponding cis configuration, but not physically as small as the 7 nm conductance-derived diameter would suggest (vide supra).²⁰ We fabricated unfunctionalized, readily wetting, CBD ${\text{SiN}}_{\mathrm{x}}$ nanopores where Equation (1) can be used to calculate a physical nanopore diameter—here, at 5 nm smaller than even the effective diameter of the trans nanopore. We could readily detect both maltodextrin and DNA in these small nanopores (Figures S15 and S16), thus elevating the influence of the trans nanopore hydrophobicity on the sensing selectivity. To investigate this, we added an organic solvent, acetonitrile, to our electrolyte and attempted to sense both molecules in this new medium. We chose acetonitrile because the free trans 4-(propargyloxy)azobenzene (with alkene terminus) was soluble in acetonitrile at room temperature when it was insoluble in electrolyte (Figure S8). We were able to detect discrete events in the presence of both analytes at the analyte-specific voltage polarities corresponding to detection in Figure 2, thereby giving strong indications of the importance of nanopore coating hydrophobicity in an aqueous medium in determining the sensing selectivity. Electrolyte composition thus emerges as an additional potential parameter for tuning nanopore selectivity. Photoswitchable translocation in the photochromic nanopores can be useful as a molecular delivery system. For biopolymer data storage schemes, photo-switching can turn the detector on or off and voltage determines which biopolymers will be read out. Multiplexed data storage using two different biopolymers can increase the information storage density and the same photochromic pore can be used for readout with voltage polarity used to select the polymer of choice.

CONCLUSIONS

We have successfully created a metastable, stimulus-responsive nanopore using a conventional thin-film material used at scale in consumer microelectronic devices. The robust photoswitch continued to function even under confinement inside the pore, after measurements, and across weeks. The pore could be reversibly switched over repeated cycles between a smaller, lower-conductance state and a larger, higher-conductance state accompanied by other changes such as to the surface coating polarity. These pores are practicable for applications even at the level of geometric aperture (e.g. high throughput chemically-tuned filtering) and ionic circuit element (e.g. as photoswitched conductor and digital bit). The more striking application vistas emerge from the ability to use light to switch between on/off single-molecule responsive states of the nanopore, and the ability to use voltage polarity to switch between biopolymer class (foreshadowing multiplexed molecular data storage possibilities). The proof-of-principle results for photoswitchable DNA sensing point to opportunities in exploring how to tailor the photochromic coating—by derivatization or class—to optimize both nanopore sensing performance against clogging, for example, and switching capability. Single-molecule sensing results for the glycan maltodextrin were comparable to those using a conventional ${\text{SiN}}_{\mathrm{x}}$ nanopore when photoswitched from the nonresponsive state, speaking to the general utility of this photochromic solid-state pore. We have, in sum, developed a thin-film platform—with diverse control parameters—for studying, tuning, and harnessing photoresponsive mass transport in diverse domains from filtration to single-molecule sensing of importance to the grand visions of information storage, genomics, and glycomics.

METHODS

Nanopores were formed by controlled (dielectric) breakdown (CBD)^{23, 61} in 15 nm-thick ${\text{SiN}}_{\mathrm{x}}$ membranes and then functionalized with custom synthesized 4-(propargyloxy)azobenzene by photohydrosilylation in acetonitrile (Scheme 1)²⁰. All characterizations of and measurements using the photochromic nanopores were performed using a Teflon housing inside a Faraday cage that blocked ambient light and in conventional 1 M KCl aqueous electrolyte buffered to pH 7 with 10 mM HEPES. Except where noted, all nanopore photochromic switching experiments were performed on functionalized nanopores that had been removed from their Teflon housing and immersed in acetonitrile. All current measurements were conducted in the dark. More detailed descriptions of experimental methods and materials are provided in the supporting information and in earlier sources.^{24, 55}