The details of the chemicals and instruments used are provided in the Supplementary Text 1.

The PDMS slices were positioned on a metal table and treated with a commercial 10.6 μm CO₂ laser (MSDB-FM60, Minsheng Laser, China, 60 W) in ambient air. The parameters were set at a speed of 1,000 mm/s, a power of 1.8 W, a frequency of 10 kHz, and a pulse density of 5 pulses per dot, henceforth referred to as DLW_1. DLW_n denotes n repetitions of a laser scan under the DLW_1 condition. After treatment, the samples were sectioned into 1 cm × 1 cm pieces for further experiments.

The chemical bonding in DLW-PDMS was investigated using attenuated total reflectance-Fourier transform infrared spectroscopy (ATR-FTIR) (Bruker, Germany) over the range 500-3,200 cm^-1. A Raman microscope (inVia, Renishaw, UK) equipped with a 532 nm excitation was also employed. The exposure conditions were set to 10% power for 10 s under a 20× lens. The X-ray diffraction (XRD) pattern was obtained using a Rigaku instrument (SmartLab, Japan) with Cu Kα radiation. Each sample was scanned within the 10°-60° range at a rate of 2°/min. X-ray photoelectron spectroscopy (XPS) spectra were collected using a Thermo Scientific K-Alpha spectrometer (USA) operated at an accelerating voltage of 12 kV. Monochromatic Al Kα radiation was employed as the X-ray source. The surface structure of DLW-PDMS was visualized using a Hitachi Regulus 8100 scanning electron microscopy (SEM, Japan). Hydrophobicity was quantified using a DataPhysics OCA 15EC instrument (Germany). A 5 μL deionized-water droplet was deposited onto the sample surface to measure the water contact angle (WCA) and sliding angle (SA). For the SA measurement, the stage was conducted at a tilt speed of 1.0 °/s. Data represent average values from three independent samples per group, with three spots measured on each sample.

E. coli (ATCC 25922) was prepared in LB broth, resuspended in 1× phosphate-buffered saline (PBS) solution, and adjusted to 10⁸ colony-forming units per mL (CFU/mL). The DLW-PDMS samples were submerged in the bacterial solution at 37 °C for 24 h. The slices were then removed, rinsed gently with 1× PBS, and transferred to centrifuge tubes containing 5 mL of buffer for a sequential vortex-sonicate-vortex treatment (2-4-2 min). The bacteria were then counted using the plate counting method.

Bacterial viability was evaluated using a live/dead kit (Thermo Fisher, USA) and a microreader (SpectraMax iD3, Molecular Devices, USA). The image was captured with a 40× confocal fluorescence microscope (CLSM, Nikon, Japan). For long-term antifouling assessments, a working solution of E. coli expressing green fluorescent protein (E. coli-GFP, ATCC 25922-GFP) was cultured with the DLW-PDMS sheets at 37 °C for 7 days. Following rinsing, surface biofilm formation was captured using CLSM. The excitation wavelength was 488 nm, with a Z-stacking interval of 2.5 μm. The 3D visualizations were generated using the 3D viewer plugin in FIJI^[20].

The preparation of the DLW-PDMS microchannel is detailed in Supplementary Text 3. The treated chips underwent a triple ethanol-deionized (DI) water cleaning before evaluation. The chips were positioned on a heating table at 37 °C. A 50 mL syringe containing a bacterial solution of E. coli-GFP at 10⁸ CFU/mL was circulated through the DLW-PDMS chip at 5 μL/min, with the solution renewed every 48 h to preserve bacterial activity. Following a week-long circulation of the bacterial solution, unattached bacteria were cleared with a PBS rinse, and the chips were then observed at 20× magnification using CLSM.

Electrochemical impedance spectroscopy (EIS) and cyclic voltammetry (CV) experiments were utilized to examine the electrochemical properties of DLW-PDMS as a working electrode. EIS experiments were conducted on a Zahner electrochemical workstation (Zennium XC, Germany) using a 100 mM solution of [Fe(CN)₆]^3-/4- in 0.1 M KCl. The experimental conditions were set to an amplitude of 10 mV and a frequency range of 0.1-100,000 Hz. The CV experiments were performed in a 5 mM [Fe(CN)₆]^3-/4- containing 0.1 M KCl. Routine CV characterization was conducted at a scan rate of 50 mV/s over the range of -0.6 to 1.2 V. For parameter-specific measurements, CV at 30 mV/s from -0.1 to 0.5 V was used to calculate the electron transfer rate constant (k⁰) and CVs at varied scan rates were recorded to determine electrochemically active surface area (A). For double-layer capacitance (C_dl), CVs at varied scan rates were acquired in PBS from 0.2 to 0.4 V. Independent scans from -1.2 to 2.0 V were used to evaluate the electrochemical window (ECW). Detailed calculations are available in the Supplementary Materials.

The electrochemical polymerization was carried out in a 10 mL solution containing 0.1 M KCl, 0.05 M pyrrole, and E. coli at 10⁸ CFU/mL. The process was performed within an ECW of -0.3 to 0.7 V at a scan rate of 50 mV/s. After eight electrochemical cycles, the sample was transferred to an elution solution containing 5% w/v sodium dodecyl sulfate (SDS) in 5% v/v HAc solution (pH = 2.47) and immersed for 1 h to remove the bacteria. Following the elution step, the electrode was gently rinsed with sterilized DI water. The resulting electrode was identified as the BIP-DLW-PDMS sensor. A non-imprinted polymer (NIP) electrode was prepared using the same steps without adding bacteria during electropolymerization and served as a control. SEM was employed to study the morphology of each step in the process, while E. coli-GFP was used for in-situ confocal fluorescence imaging (60× lens).

The prepared BIP-DLW-PDMS was incubated with various concentrations of E. coli from 10³ to 10⁸ CFU/mL for 2.5 h. EIS measurement was conducted in a 5 mM [Fe(CN)₆]^3-/4- solution containing 0.1 M KCl over a frequency range of 0.1 to 100,000 Hz. The impedimetric changes were processed using: ΔR_ct = |R_ct1 - R_ct0|, where R_ct1 is the value of the R_ct of the BIP-DLW-PDMS sensor with bacteria, R_ct0 is the value without bacteria.

S. aureus (ATCC 25923), S. enterica (ATCC 13076), P. aeruginosa (ATCC 27853), and E. faecalis (ATCC 6569) were selected as interferents [Supplementary Table 1]. Individual measurements for all tested bacteria, including the target E. coli, were performed at 10⁸ CFU/mL. For mixed-culture competitive selectivity tests, E. coli (10⁷ CFU/mL) was mixed with pooled non-target bacteria (total 10⁸ CFU/mL), corresponding to a target-to-nontarget concentration ratio of 1:10.

The real samples, including tap water, seawater, and milk, were collected from local sources. Milk samples were pretreated by a two-fold dilution before measurement. The real samples were spiked with 10⁶ CFU/mL E. coli and tested in triplicate.

Initial single-pass DLW treatments under varied parameters [Supplementary Table 2 and Supplementary Movie 1] produced surfaces with minor visual differences [Supplementary Figure 1]. Still, they induced significant compositional changes in the Si-O, Si-C, and carbon [Supplementary Figure 2]. Owing to PDMS’s substantial infrared absorption at 10.6 μm, a CO₂ laser at this wavelength enables efficient photothermal modification under ambient-air conditions^[21]. Notably, increased carbon content and reduced SiO₂ correlated with improved antifouling efficacy [Supplementary Figure 3]. This observation prompted a strategy of increased laser-overlap exposures, hypothesized to promote a more ordered, graphitic carbon structure^[22], which is known to exhibit superior interfacial properties compared to amorphous carbon^[23,24].

To investigate this hypothesis, the PDMS substrate was subjected to multiple laser scans under the optimal antifouling parameter condition identified by principal component analysis (PCA), denoted F10k [Supplementary Figure 4]. This repetitive process induced a distinct visual transformation from white to black [Supplementary Figure 5], indicative of substantial photothermal modification. FTIR spectra revealed a surface chemistry dominated by Si–O and Si–C bonds [Figure 1A]. A progressive narrowing of the Si–O peak [Supplementary Table 3] suggests the selective ablation of the amorphous SiO₂ layer with repeated laser exposure^[25]. Concurrently, Raman spectroscopy demonstrated a pronounced increase in the D and G band intensities [Figure 1B]. The intensity ratio (I_D/I_G) decreased with successive scans, reaching 1.13 for the DLW_3 condition (DLW_n denotes n laser-scan repetitions), consistent with the formation of an ordered, pyrolyzed carbon structure^[26]. Extended scanning further weakened both carbon-related bands and raised the I_D/I_G ratio to 1.23, reflecting structural evolution and partial degradation of the carbon phase.

XRD analysis provided further insight into the crystalline evolution of the surface [Figure 1C]. The untreated PDMS exhibited broad amorphous peaks (2θ = 12° and 22.6°). However, the laser-treated surfaces initially showed one broad peak near 23°, characteristic of amorphous SiO₂ or SiOC^[27,28]. With increasing numbers of laser repetitions, distinct diffraction peaks corresponding to crystalline SiC and graphitic carbon emerged^[29,30]. These signals imply phase segregation during the formation of crystalline domains, consistent with previously reported mechanistic analyses^[31,32]. Notably, the absence of characteristic SiC diffraction peaks in DLW_4 agrees with the trends observed in Raman measurements and with earlier reports that excessive thermal exposure drives carbon vaporization and suppresses SiC formation^[28].

To elucidate the precise surface chemistry, XPS was performed on the DLW samples [Figure 1D and Supplementary Figure 6]. Deconvolution of the high-resolution C 1s spectrum resolved peaks at sp²-C (284 eV), Si-C (284.7 eV), sp³-C (286 eV), and C=O (288.3 eV), while the Si 2p region indicated the presence of Si-C (101.5 eV) and a combined Si-O/Si-OC phase (102.9 eV)^[33,34]. The measured elemental composition for the DLW_3 surface (Si = 26.81 at%, O = 34.44 at%, C = 38.75 at%; Supplementary Table 4) is in close agreement with values reported for pure SiOC^[34]. The reduction in carbon content for DLW_4 (33.71 at%, Supplementary Table 4) coincides with structural changes detected by Raman and XRD measurements, suggesting carbon depletion under excessive laser irradiation. These results revealed that repetitive laser pyrolysis in ambient air transforms the PDMS surface into a composite material comprising an amorphous SiOC matrix embedded with crystalline SiC and graphene-like carbon nanoparticles^[17,35].

The laser-induced physicochemical transformation culminated in a hierarchical surface morphology, characterized by a wrinkled, multi-scale structure, as observed by SEM [Figure 1E and Supplementary Figure 7]. The resulting micro- and nanoscale topography, combined with the altered surface chemistry, drove a transition to superhydrophobicity. The WCA increased from 112° on pristine PDMS to 157° for the DLW_3 condition [Figure 1F], accompanied by a dramatic reduction in the SA to 3° [Figure 1G and Supplementary Figure 8], indicating a robust Cassie-Baxter wetting state consistent with a previous report^[15]. Further increases in the number of laser scans produced heavily wrinkled surface topography for DLW_4. These morphological alterations, together with the compositional degradation identified above by Raman, XPS, and XRD, coincided with degraded wetting performance: the WCA decreased to 139°, whereas the SA increased to 11°. The combined set of observations points to partial volatilization of surface-enriched SiOC/C and disruption of the beneficial hierarchical micro-nanotexture, counteracting the advantages achieved under moderate repeated laser exposure.

The antifouling efficacy of the DLW-PDMS surfaces was systematically evaluated via CFU assays. The DLW_3 surface exhibited optimal performance, achieving a 94.47% reduction in E. coli attachment after 24 h compared to the untreated PDMS control (P = 0.0019, Figure 2A). Consistently, confocal images revealed synchronous reductions in both live and dead bacteria [Supplementary Figure 9]. This efficiency is highly competitive with state-of-the-art LIG filtration membranes loaded with 0.5 mg/cm² graphene oxide (GO, 95%)^[36] and PDMS doped with gallium-based liquid metal nanodroplets (90%)^[4].

When moving to the DLW_4 condition, antifouling performance declined. Despite observing a higher ratio of dead attached bacteria on DLW_4 [Supplementary Figure 9F], concurrent increases in both live and dead bacterial and CFU counts do not support a bactericidal effect. The elevated dead-cell proportion likely arises from prolonged residence of bacteria trapped in surface crevices after adhesion, rather than from direct membrane rupture induced by surface topography^[37].

Long-term stability assessments under static culture revealed extensive biofilm formation on the pristine PDMS surface. In contrast, the DLW_3 surface supported only sparse, vertically oriented microcolonies [Figure 2B]. This profound resistance to biofilm development is driven by the hierarchical topography, which stabilizes an interfacial air layer, thereby minimizing the solid-liquid contact area available for bacterial anchoring^[38].

Recent research highlights laser-engineered topographical patterning as an effective strategy for constructing cell-repellent biointerfaces for implantable bioelectronics, in which surface structuring suppresses cell adhesion without relying on biocidal components^[39]. Moreover, prior literature indicates that SiOC particles exhibit favorable biocompatibility and exert no inhibitory effect on bacterial proliferation^[40-42]. These findings support our inference of a predominantly anti-adhesive mechanism.

To demonstrate the practical utility of this technology for mitigating biofouling, a critical challenge in PDMS-based microfluidics, the DLW process was applied to the interior of microscale channels [Figure 2C]. The modified channels were assembled into microfluidic devices and subjected to 7 days of continuous circulation with a high-load bacterial suspension (7 days with the wall shear stress only at 0.25 Pa, Supplementary Text 3). Post-experiment analysis revealed substantial bacterial accumulation and biofilm formation within the untreated PDMS control chip, particularly along the channel edges [Figure 2D]. Conversely, the DLW_3-treated channel exhibited only minimal bacterial adhesion, representing a statistically significant reduction in surface fouling (P = 0.027). These results validate the mechanical robustness and operational stability of the superhydrophobic SiOC layer under dynamic flow conditions.

Due to the hierarchical, superhydrophobic nature of the DLW_3 composite (WCA = 157°), the thermodynamic barrier to biofouling was evaluated via the apparent work of adhesion (W_a). Applying the Young-Dupré equation: W_a = γ_w(1 + cosθ_w), the W_a of the optimized SiOC matrix was calculated to be exceptionally low (≈ 5.8 mN/m). This thermodynamic barrier to aqueous wetting disfavors the stable anchoring of E. coli extracellular polymeric substances, which could somewhat explain the observed reduction in bacterial adhesion. Nevertheless, it should be noted that the adhesion derived from WCA measurements serves only as an effective surface-wetting descriptor. Bacterial adhesion is governed by multiple additional factors beyond surface wettability, including hydration state, extracellular substances, and surface charge^[37].

The formation of graphene-like carbon structures within the SiOC matrix^[43], combined with the established biocompatibility and chemical stability of polymer-derived SiOC ceramics as electrode materials^[44], motivated a rigorous investigation of the platform’s electrochemical properties. As shown in Figure 3A, the EIS plot revealed a dramatic decrease in R_ct, from 44.4 Ω for a single pass (DLW_1) to < 1 Ω (DLW_3). This value significantly outperforms comparable SiOC/C materials (R_ct = 82.78 Ω)^[45]. These findings, which position the DLW_3 surface as a highly conductive electrode, were corroborated by CV measurements [Figure 3B]. The CV response and redox currents rose with laser repetitions, peaking at DLW_3 and decreasing at DLW_4. This trend aligns with our materials characterization, which indicates that three repetitions yield an optimal balance of high graphitic carbon content and low structural defects [Figure 1 and Supplementary Table 4]. In contrast, excessive exposure may lead to the degradation or vaporization of the conductive carbon phase^[46]. This observation further supports the conclusion that DLW-PDMS forms a composite material consisting of SiOC/C. While pure SiO₂ is intrinsically insulating, the generated interconnected carbon network provides pathways for efficient electron transport across the substrate^[19,47].

Under time-varying electric fields, transient Maxwell electrostatic stress precludes instantaneous steady-state force balance, as illustrated by AC-electrowetting observations where droplet inertia hinders cycle-resolved interfacial relaxation^[48]. In the CV measurements, a continuously swept potential generates a transient Maxwell stress concentrated at the liquid-air-solid three-phase contact lines. This stress may locally overcome the Laplace capillary barrier, inducing spatially confined plastron collapse without a global Cassie-to-Wenzel transition or degradation of hierarchical micro/nanostructures^[49,50]. Unlike static-DC electrowetting, no steady-state Maxwell-Laplace equilibrium is established during potential scanning. Benefiting from a hierarchical microstructure, wetting is confined to isolated domains, and macroscopic superhydrophobicity is preserved^[51].

Quantitative analysis of the CV data [Table 1] revealed a heterogeneous electron transfer rate constant (k⁰) of 0.00708 cm/s for DLW_3, rivaling standard electrode materials such as glassy carbon (k⁰ = 0.00782 cm/s) and pyrolytic carbon (k⁰ = 0.0054 cm/s)^[52,53]. A good linear relationship between peak current and the square root of sweep rate demonstrates that the electrochemical reaction is governed by semi-infinite diffusion of redox-active species toward the electrode surface [Figure 3C and Supplementary Figure 10]. As shown in Figure 3D, the electrochemically active surface area (A) reached its maximum of 1.674 cm² for DLW_3, which is attributed to the laser-pyrolyzed hierarchical roughness.

Analysis of the non-Faradaic potential regions revealed quasi-rectangular CV curves, indicative of double-layer capacitive behavior [Supplementary Figure 11]. The DLW_3 electrode maintained this ideal rectangular shape even at high scan rates, suggesting excellent charge propagation and electrochemical stability. DLW_3 electrode’s double-layer capacitance (C_dl) reached a minimum of 60.50 μF/cm² [Figure 3E and Table 1], minimizing non-Faradaic background currents, thereby improving the signal-to-noise ratio. This value is comparable to conventional noble metal electrodes (platinum: 64 μF/cm², silver: 53 μF/cm²)^[52], and superior to other laser-modified materials, such as carbon black-infused polylactic acid (PLA, 96.2 μF/cm²)^[54]. The stable ECW of the DLW-PDMS electrode spanned -0.7 to +1.5 V [Figure 3F], a wide operational range similar to that of standard graphite (-0.5 to +1.6 V) and pyrolytic carbon (-0.9 to +1.6 V) electrodes^[52], confirming the platform’s suitability for diverse electroanalytical applications.

A defining feature of the DLW_3 platform is its ability to reconcile superhydrophobicity with high-performance electrochemistry. Traditional electrochemical theory dictates that a Cassie-Baxter state (WCA = 157°) should severely impede Faradaic processes due to the insulating air plastron. However, the DLW_3 electrode exhibits interfacial electron transfer kinetics (k⁰ = 0.00708 cm/s). We attribute this unique capability to a potential-driven electrowetting transition. While the hierarchical SiOC matrix thermodynamically repels bacterial suspensions under open-circuit conditions, preventing the initial formation of focal adhesion complexes, applying sweeping potentials during CV locally reduces the liquid-solid interfacial tension. This transient, potential-induced wetting allows the aqueous electrolyte to penetrate the graphitic nano-domains, dynamically facilitating rapid charge transfer. Crucially, upon removal of the applied potential, the robust hierarchical topography prevents total Wenzel wetting, allowing the macroscopic antifouling properties to persist under dynamic microfluidic flow. This dynamic interfacial behavior distinguishes the DLW_3 composite from conventional LIG electrodes, which become rapidly fouled, and standard superhydrophobic polymers, which remain electrochemically inert. The reproducibility of this favorable electrochemical performance was further verified by impedance measurements, which revealed negligible intra-batch variation and excellent batch-to-batch consistency for DLW_3 [Supplementary Table 5].

Repeated laser scanning produces a suite of competing effects that dictate final surface properties: progressive laser-driven carbonization generates electroactive graphitic domains embedded within the SiOC matrix; evolving multiscale surface roughness builds the hierarchical micro-nano topography essential for Cassie-Baxter superhydrophobic antifouling; cumulative laser input induces heat accumulation, thermal stress, and solid-state phase transformations originating from thermal-relaxation processes; and excessive laser fluence causes material ablation. These effects respond differently as the number of repetitions increases. Moderate repeated scanning from DLW_1 to DLW_3 yields favorable carbon enrichment and well-defined hierarchical morphology without severe ablation or thermal damage, delivering balanced antifouling and electrochemical performance. Further increasing scan passes to DLW_4 breaks this optimal trade-off: carbon content decreases, and electrochemical activity weakens owing to ablation-driven loss of graphitic carbon; meanwhile, excess heat generates thermal-stress-mediated structural defects, and ablation degrades the Cassie-Baxter-supporting hierarchical topography. These combined factors explain the superior overall performance observed for the DLW_3 condition.

While microfluidic devices offer an attractive platform for rapid bacterial diagnostics^[55,56], integrating biorecognition elements onto PDMS surfaces remains a significant bottleneck, often requiring complex, time-consuming protocols^[57,58]. To address this, we integrated BIP technology, a robust approach for creating synthetic receptors using whole cells as templates^[59], directly onto the DLW_3 electrode.

The in-situ electropolymerization of pyrrole in the presence of E. coli templates [Figure 4A] resulted in a high-impedance state (R_ct = 1.5 kΩ in Figure 4B) due to bacterial encapsulation [Figure 4B-D and Supplementary Figure 12]. SEM and confocal images showed the uniform entrapment of intact bacterial cells within the reticulated 3D Ppy network [Figure 4C(I) and D(I)]. Subsequent template elution generated highly specific, three-dimensional recognition cavities, as evidenced by a sharp decrease in impedance to 234 Ω [Figure 4B], with the formation of these imprinted voids within the polymer film directly visualized in Figure 4C(II) and D(II). Upon re-exposure of this BIP-DLW_3 sensor to the target E. coli, a significant increase in impedance was observed (Δ symbol in Figure 4B), indicative of the specific rebinding of bacteria within the complementary cavities, which obstructs charge transfer at the electrode interface. This successful capture event was directly visualized by SEM and confocal imaging, showing bacteria docked within the imprinted sites [Figure 4C(III) and D(III)]. Crucially, a NIP, prepared without the bacterial template as a control, exhibited negligible impedance change and minimal bacterial adhesion upon exposure to E. coli. This result demonstrates that the observed sensing response is due to the high-selectivity recognition conferred by the molecular imprinting process, rather than to nonspecific adsorption.

Under the optimal condition [Supplementary Figure 13], the BIP-DLW_3 sensor demonstrated a wide dynamic range for E. coli (10³ to 10⁸ CFU/mL), with an increased R_ct in a concentration-dependent manner [Figure 5A], yielding a limit of detection (LOD) of 348 CFU/mL (adjusted R² = 0.9882, Figure 5B). The fabrication protocol proved highly reproducible, with negligible response variation observed across five independent batches (3 replicate sensors per batch) when challenged by 10⁸ CFU/mL E. coli [Figure 5C].

While the BIP-DLW_3 sensor was exposed to four different non-target bacterial species at a high concentration (10⁸ CFU/mL), the impedance response was minimal. In contrast, the response to the target E. coli was 2.6- to 3.5-fold greater (P ≤ 0.0001), confirming the sensor’s high specificity [Figure 5D]. This selectivity was further validated in a competitive assay, where the sensor preferentially detected target E. coli that constituted approximately 9.1% of a mixed bacterial population (Figure 5E, concentration of target E. coli is 10⁷ CFU/mL). This high degree of selectivity is directly attributed to the specific recognition cavities, complementary in size, shape, and chemical functionality to the E. coli template, engineered into the polymer matrix during the imprinting process^[60].

Practical utility was validated in complex matrices (tap water, seawater, milk). The sensor’s response to E. coli spiked into tap water and seawater (final concentration = 10⁶ CFU/mL) closely matched the response observed in the clean PBS buffer control. While greater response variability was observed in minimally diluted milk samples, the results demonstrate the sensor’s robust performance and potential for direct application to real-world samples with minimal pretreatment [Figure 5F and Supplementary Table 6].

Compared to conventional bacterial methods such as agar plating and enzyme-linked immunosorbent assay (ELISA)^[61,62], this unified DLW-PDMS platform enables rapid (< 4 h), highly sensitive, and label-free bacterial identification. Notably, our BIP-functionalized DLW-PDMS electrode, without additional nanomaterials pre-modification, achieves a LOD comparable to that of gold and carbon electrodes that employ such nanostructured pre-treatment to enhance sensing performance when functionalized with diverse recognition moieties [Table 2].

Taking advantage of the template-replaceable nature of molecularly imprinted technology, our BIP sensor shows promise for extending detection from single-target detection of E. coli to a broader spectrum of analytes, including other bacterial pathogens as well as small-molecule targets such as antibiotics and glucose^[69]. This platform shows great promise for on-site rapid screening and long-term in-situ monitoring applications.