**3. Results and Discussion**

### *3.1. Multiple Dip-Coating of Spherical Analytes on Nanoplasmonic Sensors*

To examine the anisotropic wetting e ffect of the one-dimensional grating structure [20–22], the dipping and pulling directions of the DAS arrays were set parallel or perpendicular to the length direction of the Au strips. Figure 2 shows the SEM images of the PS bead-coated sample surfaces, depending on the direction and number of dip-coating cycles. Due to capillary forces induced on the three-phase (liquid–vapor–substrate) contact line [11–14], PS beads spontaneously assembled in the valleys between neighboring DAS structures and adsorbed on the surface by the van der Waals forces, regardless of the dip-coating direction. However, it should be noted that the density of the PS beads on the samples where the dip-coating direction was parallel to the Au strip (hereinafter referred to as par-dip sample) was much higher than that on the samples with a perpendicular dip-coating direction (hereinafter referred to as the per-dip sample). For example, in the case of a triple dip-coating, the population of PS beads of the par-dip sample was 154 in the SEM image (Figure 2f), which was approximately five times larger than that of the per-dip sample (Figure 2b). It is believed that this di fference was caused by anisotropic wettability and evaporation-induced flow near the contact line. Because the liquid contact angle in the direction parallel to the Au strip was smaller than the contact angle along the perpendicular direction, the par-dip sample with its long meniscus could secure a longer time for positioning the PS beads than the per-dip sample. In addition, the contact line with its long meniscus could induce an outward flow of the PS beads during evaporation; this phenomenon is known as the co ffee stain phenomenon [23,24]. Accordingly, deposits of self-assembled beads could accumulate much more on the par-dip sample than on the per-dip sample during the multiple dip-coating process. In the case of the par-dip sample, the valleys between neighboring DAS structures were almost filled with PS beads after five repeated dip-coating cycles, as shown in Figure 2g.

The valleys between the Au strips not only accommodate PS beads by capillary forces, but also act as hosts to strong plasmonic fields due to the interactions between adjacent strips, as was demonstrated in experimental and simulation studies [25]. Therefore, the PS beads self-assembled in the valley between the DAS arrays can induce a change in the resonance conditions of the Au strips, resulting in a spectral shift of the LSPR wavelength. Figure 3 shows the LSPR shifts of the per- and par-dip samples depending on the number of dip-coating repetitions. The LSPR peak shifts of the par-dip samples were much larger than those of the per-dip samples due to a greater number of dense bead deposits on the sample surfaces. This observation is in good agreemen<sup>t</sup> with the SEM images in Figure 2. As the LSPR peak shift was accompanied by a change in the amplitude of absorbance, as shown in Figure 3a,b, it would be possible to monitor the change in absorbance at a specific wavelength (e.g., the initial resonant wavelength) as an alternative signal reading method.

Figure 3a,c also indicates that the peak shift generated by the dip-coating was reduced after five repetitions of the dip-coating process. Once the valleys were filled with PS beads, the additional beads adsorbed on the upper sides of the DAS structures were too far away from the plasmonic fields developed in the valleys, and hence they did not significantly a ffect the resonant condition of the Au strips. For the same reason, analytes that are too small or beads that are too large compared to the dimensions of the valley are not very sensitive to DAS arrays, as demonstrated by experiments and theoretical calculations in our previous report [25]. Therefore, the dimensions of the DAS structures can be adapted to the size of a particular analyte, and the gold surface of DAS can be functionalized with antibodies for target-specific detection.

**Figure 2.** SEM images of the polystyrene (PS) bead-coated sample surfaces according to the direction and number of repeated dip-coating cycles. The dipping and pulling directions of DAS arrays were (**<sup>a</sup>**–**d**) perpendicular and (**e–h**) parallel to the direction of the length of Au strips. Samples were dip-coated (**<sup>a</sup>**,**<sup>e</sup>**) once, (**b**,**f**) three times, (**<sup>c</sup>**,**g**) five times, and (**d**,**h**) seven times. Scale bars represent 500 nm.

**Figure 3.** (**<sup>a</sup>**,**b**) Absorbance curves and (**<sup>c</sup>**,**d**) localized surface plasmon resonance (LSPR) peak shifts of PS bead-coated DAS arrays as functions of the number of dip-coating repetition cycles. The data in Figure 3a,c and Figure 3b,d were obtained from the par-dip and per-dip samples, respectively. All the data in Figure 3c,d showed statistically significant differences (P-value < 0.05), except for the first two data in Figure 3d (single and double dip coatings).

### *3.2. Assessment of Exosome Stability by Nanoplasmonic Sensors*

In addition to monitoring the adsorption of analytes, it is also possible to monitor desorption or degradation of analytes adsorbed on DAS arrays by measuring the LSPR shifts. Thus, DAS arrays in combination with the proposed multiple dip-coating method were used to assess the stability or degradability of exosomes in neutral or acidic microenvironment known to be associated with their drug delivery efficacy [17,18]. Exosomes dispersed in a phosphate-buffered solution at a concentration of 1 μg/μ<sup>L</sup> were accumulated on DAS arrays by thrice dip-coating. The dip-coated exosomes resulted in an 8 nm red shift of the LSPR peak (Supplementary Materials, Figure S1). For the case when a 10 μL drop of the same exosome sample was dried on a DAS array, the LSPR peak shift was less than 2 nm (Supplementary Materials, Figure S2a).

To examine the pH-dependent stability of exosomes, the exosome-adsorbed DAS arrays were immersed in a TRIzol reagen<sup>t</sup> and buffer solutions with pH 7.4, 5.0, and 4.0. TRIzol reagen<sup>t</sup> for complete fragmenting of exosomes was used as a control [26]. After treatment with TRIzol reagent, the LSPR peak of the exosome-adsorbed DAS array was blue-shifted (9.5 nm) and, thus, returned to its original position within 1 h (red bars in Figure 4a). The 1.5 nm blue shift from the original position may be the result of damage to the polymer substrate (PET) by the reagent. Compared with TRIzol reagent, it was confirmed that buffer solutions with pH 7.4, 5.0, and 4.0 did not make any noticeable damage on the DAS arrays and the substrate (Supplementary Materials, Figure S2b). In the buffer solution of pH 7.4, the physiological condition was stable for exosomes, and no apparent changes in the LSPR peak were observed, although the exosome-coated DAS array was immersed for up to 60 h (black bars in Figure 4a). Compared to the previous two extremely unstable (TRIzol) and stable (pH 7.4) cases, the peak shifts of the samples immersed in acidic buffer solutions (pH 4.0 and 5.0) exhibited gradual changes, as shown in Figure 4b. The gradual blue shift in the LSPR peaks indicates that the exosomes on the DAS array were partially desorbed from the sample surface or degraded under acidic conditions with an increase in the immersion time. As the desorption of exosomes in the pH 7.4 buffer solution was not very significant, exosome degradation is considered a possible cause of the blue shift of the LSPR peak in acidic buffer solutions. Petelsak et al. demonstrated that a lipid bilayer membrane is subject to greater interfacial tension under acidic conditions than under neutral conditions [27]. Therefore, in pH 4.0 and pH 5.0 buffer solutions, increased interfacial tension of the lipid membrane can reduce the stability of exosomes and result in a blue shift of the LSPR peak. In addition to testing the exosome stability demonstrated in this report, it is possible to monitor the loading and unloading of drug molecules from the exosomes under various physiological conditions that are the subjects of our further studies.

**Figure 4.** Normalized LSPR peak shifts measured by exosome-coated DAS arrays. After dip-coating thrice with exosomes, the samples were immersed in a TRIzol reagen<sup>t</sup> (red bars in (**a**)) and buffer solutions with pH 7.4 (black bars in (**a**)), 5.0 (black bars in (**b**)), and 4.0 (red bars in (**b**)). The peak shifts were divided by the initial shift values, measured after dip-coating thrice with exosomes for each sample.
