**2. Experimental Methods**

### *2.1. Nanofabrication of Double-Bent Au Strip Arrays*

The sample preparation process is illustrated in Figure 1. Initially, one-dimensional grating patterns with a period of 200 nm (1:1 line/space) and a height of 100 nm were generated on polyurethane acrylate (PUA)-coated polyethylene terephthalate (PET) substrate by UV-nanoimprint lithography. Then, a 30 nm-thick Au film was thermally evaporated at an oblique angle to the polymer nanograting. Because of the tilted deposition angle (35◦ from the surface normal direction) and a shadowing e ffect of the line grating, a double-bent Au strip (DAS) array was spontaneously formed in two steps, namely, UV-nanoimprint lithography and Au deposition. More details on the fabrication of DAS arrays can be found in our previous report, in which the refractive index sensitivity of about 210 nm refractive index unit (RIU)−<sup>1</sup> and a figure of merit (FOM) of 4.2 for DAS arrays were demonstrated [19].

**Figure 1.** Schematic illustration of sample preparation. (**a**) Double-bent Au strips (DAS) were thermally deposited at an oblique angle on UV-imprinted polymer nanograting structures. PUA and PET are polyurethane acrylate and polyethylene terephthalate, respectively. (**b**) A dip-coating process was used to assemble polystyrene beads or exosomes on the DAS sensors; the dip-coating process was repeated to accumulate analytes.

### *2.2. Active Accumulation of Spherical Analytes on DAS Arrays by Multiple Dip-Coating*

After preparation of DAS arrays, the samples were vertically dipped in a colloidal solution of polystyrene (PS) beads or exosomes and pulled out at a speed of 0.01 mm/s. A 1 wt % aqueous solution of PS beads was purchased from ThermoFisher Scientific (NanosphereTM size standard 3100A, Waltham, MA, USA). Between each coating cycle in the multiple dip-coating process, the samples were dried for 5 min in ambient atmosphere and then immersed again. After the dip-coating process was completed, the absorbance curves were measured using a UV-Vis spectrophotometer (UV-2600, Shimadzu, Kyoto, Japan) and the surfaces were observed using scanning electron microscopy (SEM, S-4800 FE-SEM, Hitachi, Tokyo, Japan).

### *2.3. Assessment of Exosome Stability*

Exosomes were isolated from human fibroblast cells grown in Dulbecco's minimal essential medium supplemented with 10 % fetal bovine serum, 100 U/mL penicillin/streptomycin at 37 ◦C. The supernatant was centrifuged at 800× *g* for 5 min and 2000× *g* for 10 min to remove cellular debris. The ultracentrifugation step was conducted in cell debris-removed supernatant at 100,000× *g* for 3 h to collect the pellet (Beckman, Brea, CA, USA). The exosomal protein concentration of 1 μg/μ<sup>L</sup> was measured using a Pierce BCA Protein Assay Kit. To examine the pH-dependent stability of exosomes, the exosome-adsorbed DAS arrays were immersed in a TRIzol reagen<sup>t</sup> (Invitrogen, Waltham, MA, USA), 1× phosphate buffer solution with pH 7.4 (Merck, Darmstadt, Germany), and citric acid/sodium hydroxide buffer solutions with pH 4.0 and 5.0 (Merck, Darmstadt, Germany).
