Ultrafast Plasmonic Nanocavity Thermocycler for Direct Real-Time Reverse Transcription PCR at a Point-of-Care Level

Abstract

Rapid molecular diagnostics remain constrained by two fundamental bottlenecks: nucleic acid extraction and slow thermocycling, which together limit true point-of-care deployment. Here we report ultrafast plasmonic nanocavity thermocycler (PNT) for direct real-time reverse transcription PCR (DRT-qPCR) at a point-of-care level. The PNT features a plasmonic nanocavity of clustered Au, silicon dioxide (SiO₂), and Au thin film, with Pt resistance temperature detector and a white light-emitting diode (WLED). Under WLED illumination, the PNT shows highly efficient photothermal conversion with rapid heating (20.4 °C/s) and cooling (7.5 °C/s) rates, completing 40 thermocycles in 269 s. The PNT is compactly integrated with a microlens array fluorescence camera and a disposable PCR cartridge into a palm-sized real-time PCR system. The DRT-qPCR completes sample preparation, reverse transcription, and PCR amplification in less than 15 min, achieving amplification efficiency comparable to a commercial benchtop PCR machine. Clinical validation with 60 SARS-CoV-2 samples demonstrates 87.1% sensitivity, 100% specificity, and 93.3% diagnostic accuracy, with all false negatives successfully detected upon retesting. The PNT-based DRT-qPCR can provide a precise and scalable solution for rapid molecular diagnostics in decentralized and time-sensitive settings across diverse infectious diseases.

Introduction

Continuous recurrence of infectious diseases drives the growing demand for high accuracy and easy accessibility in ultrafast nucleic acid-based diagnostics. , The COVID-19 pandemic exposed critical weaknesses in diagnostic infrastructure, highlighting the gap between centralized testing and decentralized detection. − Despite their short turnaround time and low cost, rapid antigen tests suffer from low sensitivity and lack quantitative capability, hindering early stage and accurate disease surveillance. − Consequently, real-time polymerase chain reaction (PCR) remains the gold standard for the diagnosis of infectious diseases due to its high sensitivity and specificity in pathogen detection. − However, conventional real-time PCR is time-consuming, labor-intensive, and bulky, thus limiting its use in point-of-care (POC) settings.

Commercial Peltier-based PCR systems often struggle to achieve high heating and cooling rates during thermocycling. Alternative methods such as induction, acoustic, , microwave, , and Joule heating − improve thermocycling efficiency but encounter challenges in thermal uniformity, contamination, and compact integration. Metallic nanomaterials exhibit highly efficient light-to-heat conversion through the plasmonic photothermal effect, allowing for ultrafast PCR. Gold (Au) nanoparticles such as nanoshells , or nanorods provide volumetric heating in solution but require excessive nanoparticle concentrations and high-power light sources to achieve sufficient heating. , In contrast, plasmonic nanostructures enhance thermal uniformity over a large-area substrate and photothermal conversion through strong light absorption. For instance, Au thin films absorb sub-540 nm light, inducing rapid and uniform heat under blue light emitting diode (LED) illumination. , Plasmonic nanopillar arrays achieve broadband visible absorption via localized surface plasmon resonance (LSPR), resulting in ultrafast thermocycling under white LED (WLED) illumination. − Metal–insulator–metal (MIM) nanocavities exhibit near-perfect absorption across the ultraviolet to near-infrared spectrum due to the electromagnetic field confinement, rendering them well-suited for photothermal applications. However, early use of MIM structure in PCR faces challenges such as long reaction times and lack of real-time temperature and fluorescence monitoring. Moreover, such plasmonic photothermal PCR platforms still face practical barriers from complex nanofabrication, labor-intensive sample preparation, and limited clinical validation, restricting reliable on-site molecular testing (Table S1).

Here we report an ultrafast plasmonic nanocavity thermocycler (PNT) for direct real-time reverse transcription PCR (DRT-qPCR) for point-of-care molecular diagnostic applications (Figure a). The PNT features a stacked arrangement of clustered Au–insulator–metal (cAu-IM) nanocavity with integrated platinum resistance temperature detector (Pt-RTD) on a single WLED module. The cAu-IM nanocavity consists of clustered Au (cAu), a silicon dioxide (SiO₂) layer, and an Au thin film. cAu exhibits broadband absorption due to the electromagnetic field enhancement from Au clusters with different sizes. The SiO₂ spacer induces Fabry–Pérot resonance between light transmitted through cAu and reflected from the Au film. Consequently, the PNT achieves high photothermal efficiency for ultrafast thermocycling by confining white light within the plasmonic nanocavity under modulated WLED illumination. The PNT integrates a fluorescence camera and a disposable PCR cartridge in a palm-sized configuration, realizing extraction-free, rapid point-of-care molecular diagnostics.

1.

Plasmonic nanocavity thermocycler (PNT) for ultrafast direct reverse transcription quantitative polymerase chain reaction (DRT-qPCR) at a point-of-care level. (a) A schematic illustration of fully integrated thermocycler for direct (extraction-free) RT-qPCR within a total turnaround time of 15 min. PNT consists of platinum resistance temperature detector (Pt-RTD) on clustered Au–insulator–metal (cAu-IM). The nanocavity confines visible light, producing high absorption and a robust plasmonic photothermal effect for PCR thermocycling, modulated by on–off operation of a white LED (WLED). (b) Wafer-scale nanofabrication steps of PNT, simply using thermal evaporation of Au and plasma-enhanced chemical vapor deposition (PECVD) of SiO₂, followed by the integration of Pt-RTD with hydrogen silsesquioxane (HSQ) passivation. (c) Scanning electron microscope (SEM) images of clustered Au (cAu) and cross-sectional cAu-IM nanostructure. (d) Top and bottom optical images of PNT. The left displays Pt-RTD on the top side, while the right demonstrates the substantial light absorption of the bottom side of PNT under WLED illumination.

Results and Discussion

Wafer-Level Nanofabrication of Plasmonic Nanocavity Thermocycler

The PNT is fabricated at a wafer scale, integrating Pt-RTD on the cAu-IM nanocavity (Figure b). First, cAu is formed by thermal evaporation of 8 nm thick Au onto a borosilicate glass (BSG) substrate. No adhesion layer is introduced between the BSG substrate and the cAu layer to preserve the Au cluster morphology. A 1 nm thick chromium (Cr) adhesion layer is deposited on the cAu by electron-beam evaporation, followed by deposition of an 80 nm thick SiO₂ layer using plasma-enhanced chemical vapor deposition (PECVD). A 2 nm thick Cr adhesion layer and a 30 nm thick Au thin film are then sequentially deposited on the SiO₂ layer by electron-beam evaporation and thermal evaporation, respectively, completing the cAu-IM nanocavity structure. The cAu-IM is further coated with a 2 nm thick Cr adhesion layer by electron-beam evaporation and passivated with a 500 nm thick SiO₂ layer using PECVD. A 5 nm thick Cr adhesion layer and a 100 nm thick platinum (Pt) layer are subsequently deposited by electron-beam evaporation. The Pt-RTD is patterned using photolithography and lift-off, then annealed at 200 °C for 30 min to achieve uniform sheet resistance. Final passivation is applied with a 500 nm thick hydrogen silsesquioxane (HSQ) resist layer to ensure electrical isolation from the PCR cartridge. The scanning electron microscope (SEM) clearly shows a top view image of the cAu with narrow nanogaps and a cross-sectional image of the cAu-IM nanocavity (Figure c). The optical image displays the fully fabricated PNT, featuring the Pt-RTD on cAu-IM (12 mm × 30 mm) (Figure d and S1a).

Broadband Light Absorption of cAu-IM Nanocavity

Figure illustrates the broadband light absorption of cAu-IM structure, which is crucial for the plasmonic photothermal effect. The cAu formation follows the Volmer–Weber growth mode, where thermally deposited Au adatoms nucleate into randomly sized clusters with nanoscale gaps (Figures a and S1b). The SEM images demonstrate the morphological transition from isolated Au clusters with nanogaps to percolated networks, finally forming an Au thin film as Au thickness increases. Such nanogaps in cAu function as plasmonic hotspots, thus enhancing local electromagnetic field confinement for strong light absorption, as shown by the finite-difference time-domain (FDTD) calculation (Figures b and S2). The measured absorption spectrum indicates that Au thin films (>10 nm) mainly absorb in the 400–550 nm range due to interband transitions (Figures c and S3a–d). In contrast, the cAu in ultrathin Au films (≤10 nm) features abundant nanogaps, supporting strong localized surface plasmon resonance in the 550–700 nm range. In particular, 8 nm thick cAu achieves a broad and strong plasmonic response across the visible range. The cAu-IM structure, which incorporates an 80 nm SiO₂ spacer between an 8 nm thick cAu layer and a 30 nm thick Au thin film, further enhances absorption across the full visible range (Figures d and S3e–h). The SiO₂ layer functions as a dielectric spacer, forming a Fabry–Pérot cavity where incident light is partially absorbed by the Au thin film, while the remaining light undergoes multiple reflections between the cAu layer and the Au mirror, reinforcing cavity modes (Figure S4). The 80 nm SiO₂ spacer induces the first-order Fabry–Pérot resonance, creating constructive interference and enabling the formation of multiple resonant modes. This overlap of Fabry–Pérot resonance, combined with the broadband localized surface plasmon resonance of the cAu layer, results in enhanced light trapping and broadens absorption across the full visible spectrum. Consequently, the combined effect of cAu LSPR and Fabry–Pérot resonance significantly boosts WLED absorption efficiency from 37.3% (cAu) to 81% (cAu-IM) (Figure S5).

2.

Broadband light absorption of clustered Au–insulator–metal thin film structure (cAu-IM) for plasmonic photothermal effect. (a) Nanoscopic growth of cAu depending on the thickness of Au thin film. SEM images show the evolution from isolated nanoislands to near-continuous films with increasing Au thickness, characteristic of Volmer–Weber growth. (b) Calculated electric field distribution of cAu using the finite difference time domain (FDTD) method. (c) Measured absorption spectrum (upper) and calculated absorption for WLED (lower) of cAu depending on the evaporation thickness of Au thin film. (d) Measured absorption spectrum (upper) and calculated absorption of cAu-IM for WLED (lower) depending on the insulator thickness. cAu-IM with 80 nm thick SiO₂ (red) exhibits maximum absorption across the entire visible spectrum compared to the 40 nm thick Au thin film (dashed). A strong correlation is observed with absorption profiles calculated via FDTD.

Photothermal Performance for Ultrafast Thermocycling

Photothermal performance of the cAu-IM is experimentally validated for ultrafast thermocycling applications. The photothermal enhancement is compared with a 40 nm Au thin film under single-color LED illumination at 477, 527, 539, 593, and 634 nm (Note S1 and Figure S6). The 40 nm Au thin film is selected as a mass-matched planar control because the cAu-IM structure contains approximately 38 nm of total nominal Au thickness, consisting of an 8 nm lower cAu layer and a 30 nm upper Au film. The cAu-IM clearly exhibits higher efficiency of photothermal temperature enhancement across all wavelengths due to strong and broadband light absorption (Figure a and Table S2). The maximum achievable temperatures under constant WLED illumination are compared to further examine photothermal heating performance (Figure b). The cAu-IM reaches a peak temperature of approximately 180 °C, which is 72.3 °C higher than that of the Au thin film under identical illumination conditions. This substantial difference also exhibits the notable photothermal conversion efficiency of the cAu-IM structure. The photothermal temperature profile demonstrates rapid and stable two-step thermocycling between 60 and 95 °C driven by WLED on–off modulation (Figures c and S7). The cAu-IM completes 10 thermal cycles within 94 s, whereas the Au thin film requires 176 s under identical conditions, indicating limited thermal responsiveness due to lower photothermal conversion efficiency. The performance of the PNT (cAu-IM integrated with Pt-RTD) demonstrates its capability for rapid thermal cycling (Figures d and S8). The PNT exhibits high ramp rates of 20.4 °C/s for heating and 7.5 °C/s for cooling, which are higher than those of the Au thin film (10.6 °C/s and 5.7 °C/s, respectively). The rapid heating and cooling rates result in the completion of 40 thermal cycles within 269 s (less than 4.5 min), whereas the Au film-based system requires 394 s under the same conditions. The improved cycling performance is attributed to the efficient photothermal conversion of the cAu-IM.

3.

Photothermal effect of cAu-IM based plasmonic nanocavity thermocycler (PNT). (a) Photothermal temperature enhancement (ΔT) efficiency and absorption of cAu-IM and Au thin film for different single-color LED illumination. (b) Maximum temperatures of cAu-IM and Au thin film under constant WLED illumination. (c) Photothermal temperature profiles of WLED-driven two-step thermocycling between 60 and 95 °C for 10 cycles. (d) Performance of the PNT (cAu-IM integrated with Pt-RTD), completing 40 cycles in 4.5 min via fast heating and cooling rates.

Quantitative Amplification of SARS-CoV-2 Plasmid

Rapid amplification and quantitative detection of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) are validated using a custom-built, hand-held RT-qPCR system. The platform integrates the PNT, an ultrathin microlens array fluorescence camera, and a disposable PCR cartridge (Figure S10). The PCR mixture (primers, FAM-labeled TaqMan probe, enzyme mix, reaction buffer, bovine serum albumin [BSA], and template) is loaded into the 2 μL chamber of the cartridge (Table S3). Infrared thermal imaging confirms spatially uniform temperature distribution within the 2 μL reaction chamber during PNT operation (Figure S11). The assay targets the N1 region of the SARS-CoV-2 nucleocapsid (N) gene, amplifying a 72 bp sequence (Figure a). Amplification curves obtained from plasmonic two-step qPCR demonstrate successful detection of the SARS-CoV-2 N gene plasmid within the tested quantitative range of 10²–10⁴ copies/μL (Figure b). The fluorescence signals exhibit exponential growth during cycling, with representative images captured at 5-cycle intervals illustrating real-time amplification progression. The standard curve shows a strong linear correlation between the cycle threshold (C_t) and logarithmic template concentration, supporting robust quantification performance (Figure c). Plasmonic PCR with a 2 μL reaction volume achieves an amplification efficiency of 98.2%, comparable to benchtop PCR and within the accepted range (90–110%) for qPCR. The C_t values are delayed by approximately 3.1 cycles compared to the 20 μL benchtop reactions, consistent with the expected shift resulting from a 10-fold lower total template input at equal concentration (Table S4). When performed at the same reaction volume (2 μL), benchtop PCR yields C_t values within 0.3 cycles of those obtained using the plasmonic system, supporting that the primary cause of the C_t shift is the reduced reaction volume. Note that increasing reaction volume may mitigate C_t delay in C_t-sensitive applications. Gel electrophoresis further verifies the specific amplification of the 72 bp target fragment without observable nonspecific byproducts (Figure d). The relatively weaker band intensity in the plasmonic PCR lane is attributed mainly to the smaller total reaction volume and lower total amplicon mass generated from the 2 μL plasmonic reaction compared with the 20 μL benchtop reaction. Preliminary tests are conducted to assess analytical specificity and classification performance, using SARS-CoV-2 N gene plasmid (n = 9) and no-template controls (NTCs, n = 4) (Figure e). All SARS-CoV-2 N gene plasmid yield C_t values below the cutoff of 40, whereas no amplification is observed in NTCs, ensuring clear positive–negative discrimination. As a result, receiver operating characteristic (ROC) curve analysis yields an area under the curve (AUC) of 1.0, indicating excellent classification performance with high sensitivity and specificity (Figure f).

4.

Amplification and quantification of SARS-CoV-2 N gene plasmid using ultrafast plasmonic real-time PCR. (a) Schematic of the SARS-CoV-2 genome highlighting the 72 bp N1 target region in the N gene used for PCR. (b) Amplification curves from two-step PCR (50 cycles) depending on the concentration of SARS-CoV-2 N gene plasmid. The bottom figures show fluorescence images during PCR, taken every 5 cycles. (c) Standard curve of the cycle threshold (C_t) values plotted against the concentration of the SARS-CoV-2 N gene plasmid for 2 μL in plasmonic PCR, and for 2 and 20 μL sample volumes in benchtop PCR. The low-volume plasmonic PCR demonstrates high amplification efficiency of 98.2%. (d) Agarose gel electrophoresis confirms clear 72 bp amplicon bands following PCR. (e) Preliminary test results for SARS-CoV-2 N gene plasmid and no-template controls (NTCs), establishing a C_t cutoff of 40. (f) Receiver operating characteristic (ROC) curve comparing amplification of the SARS-CoV-2 N gene plasmid and NTCs. The area under the curve (AUC) is 1.0, with a C_t cutoff of 40 cycles.

Clinical Validation of Plasmonic DRT-qPCR for COVID-19

Clinical validation of the plasmonic RT-qPCR is performed using an extraction-free protocol applied to nasopharyngeal swab specimens collected in viral transport medium (VTM) (Figure a). The samples are directly added to the PCR master mix without RNA extraction and loaded into disposable cartridges. Each cartridge is inserted into the plasmonic RT-qPCR system for reverse transcription at 50 °C (2 min), followed by two-step cycling between 91 and 60 °C. The extraction-free workflow benefits from direct dilution of specimens into the PCR mixture, thermal exposure during RT-qPCR, and the inhibitor-tolerant DRT-qPCR reagent formulation including RNase inhibitor and BSA. − The extraction-free DRT-qPCR protocol shows only a modest C_t difference of 0.78 compared with conventional extraction-based RT-qPCR using a reference material (Table S5). Fluorescence images are consecutively acquired for real-time quantification of nucleic acid amplification. The total turnaround time (TAT), including simplified sample preparation and ultrafast thermocycling, is within 15 min, enabling rapid point-of-care testing. A total of 60 clinical samples, comprising 31 COVID-19-positive and 29 negative specimens, are tested (Table S6). The resulting C_t heatmap shows a clear distinction between positive and negative samples, with specific amplification of the SARS-CoV-2 N1 target observed only in positive cases (Figure b). Using a C_t cutoff value of 40, the initial classification yields 27 true positives (TP), 29 true negatives (TN), 4 false negatives (FN), and no false positives (FP) (Figure c and Table S7). ROC curve analysis of the initial classification produces an AUC of 0.933, with 87.1% sensitivity, 100% specificity, and 93.3% accuracy (Figure d). All four initially false-negative samples produce successful amplification upon retesting (Table S8 and Figure S12), confirming the amplification capability. Retesting served only to investigate discordant samples, not as part of the standard workflow. Note that four false-negative samples result from imperfect cartridge–PNT thermal contact under the vertical operating configuration used for bubble management (Figure S13a). In this configuration, air bubbles migrate toward the upper region of the chamber and are excluded from the real-time fluorescence analysis region, while the vertical orientation can introduce local contact variation between the cartridge and the PNT surface (Figure S13b–d). The cartridge chamber protrudes slightly toward the PNT surface to improve thermal coupling, and future system refinement will incorporate a compact mechanical jig to ensure reproducible contact pressure (Figure S13e). Overall, the clinical validation results support the practical feasibility of the plasmonic DRT-qPCR platform for rapid point-of-care molecular diagnostics.

5.

Clinical validation of plasmonic direct real-time reverse transcription PCR (DRT-qPCR) for COVID-19 detection. (a) Workflow of plasmonic DRT-qPCR. Nasopharyngeal swab samples collected in viral transport medium (VTM) are directly mixed with the master mix without extraction, loaded into the cartridge, and amplified using plasmonic RT-qPCR within 15 min. (b) Heatmap of C_t values from 31 COVID-19-positive and 29 negative clinical specimens analyzed via plasmonic DRT-qPCR. (c) Diagnostic classification of true positive (TP), false negative (FN), false positive (FP), and true negative (TN) results. (d) ROC curve showing assay performance with an AUC of 0.933. The sensitivity, specificity, and accuracy are 87.1%, 100%, and 93.3%, respectively, and improve to 100% for all metrics after retesting of initial false negatives.

Conclusion

We have successfully demonstrated the plasmonic nanocavity thermocycler (PNT) for direct (extraction-free) and ultrafast reverse transcription quantitative PCR (DRT-qPCR) within a fully integrated point-of-care (POC) platform. The PNT employs a multilayer nanocavity composed of clustered Au (cAu), SiO₂, and an upper Au film, which supports broadband visible absorption through both localized surface plasmon resonance and Fabry–Pérot interference. Under white LED (WLED) illumination, the PNT achieves highly efficient photothermal conversion, producing heating and cooling rates of 20.4 °C/s and 7.5 °C/s, respectively. As a result, 40 thermal cycles are completed in under 4.5 min, nearly an order of magnitude faster than conventional Peltier-based thermocyclers, without compromising amplification efficiency or fidelity.

Integration with a disposable PCR cartridge and a microlens array fluorescence camera allows real-time monitoring of fluorescence signals and quantitative amplification in 2 μL reaction volumes with an efficiency of 98.2%, comparable to commercial benchtop systems. The entire analytical workflow, including sample preparation, reverse transcription, and amplification, is completed within 15 min through an extraction-free protocol. This simple architecture bridges the technical gap between benchtop and portable systems, advancing the practical implementation of molecular diagnostics at the point of care. Clinically, the DRT-qPCR demonstrated 93.3% diagnostic accuracy in nasopharyngeal specimens, with all initially discordant cases resolved upon retesting, underscoring both analytical reliability and operational simplicity.

Beyond speed and accuracy, the PNT framework presents a broader paradigm for accessible molecular testing. The LED-driven thermocycler eliminates bulky heat sinks, complex optics, and high-power lasers, reducing cost and power consumption while maintaining high thermal efficiency. Its compatibility with low-power electronics and plastic microfluidic cartridges enables scalable nanofabrication and use in resource-limited settings. The broadband photothermal design offers tunable light-to-heat conversion through metal–dielectric and cavity-thickness engineering. At the bioanalytical level, the same nanophotonic architecture can be further expanded toward digital partitioning, multiplex detection, and integration with upstream sample-processing modules, advancing toward fully self-contained lab-on-a-chip diagnostics. Future studies can also explore alternative photothermal materials and heterolayer absorber structures to improve economic feasibility, scalability, and application-specific photothermal performance.

Despite these advances, certain limitations warrant consideration. The small reaction volume (2 μL) that enables ultrafast thermocycling inherently limits the total number of target copies per assay, resulting in a relatively higher limit of detection compared with conventional 20 μL benchtop PCR. Optimization of reaction chamber geometry or partition-based amplification may address this trade-off between speed and sensitivity. In addition, mechanical and thermal coupling between the disposable cartridge and the PNT surface remains a potential source of variability. Improved interface design and conformal contact can further stabilize temperature distribution and ensure reproducibility. Furthermore, clinical specimens can exhibit matrix-dependent variation in viscosity, inhibitor content, and target abundance, and further validation under broader specimen conditions will strengthen the clinical translatability of the extraction-free workflow.

The PNT-based DRT-qPCR system establishes a new model for rapid and extraction-free molecular diagnostics through efficient photothermal heating and compact optical integration. By combining ultrafast plasmonic thermocycling, quantitative fluorescence imaging, and a streamlined diagnostic workflow, this approach closes the gap between benchtop precision and point-of-care usability. Future on-chip integration of sample preparation and multiplex amplification can further advance the current platform toward routine clinical and epidemiological applications, supporting rapid infectious disease surveillance and personalized molecular testing in both advanced and low-resource healthcare settings.

Experimental Section

Measurement for Absorption Spectra and WLED Absorption Efficiency

The absorption spectra (A) of plasmonic substrates were calculated using the measured transmittance (T) and reflectance (R) spectra obtained with a collimated WLED (MCWHL5-C1, Thorlabs, Inc.) coupled spectrometer (SP-2300i, Princeton Instruments, Inc.), based on the relation A = 1 − T − R. The emission spectrum of the WLED was measured using a spectrometer (C10082CAH, Hamamatsu). WLED absorption efficiency was estimated by integrating the spectral overlap between the substrate’s absorption spectrum and the WLED emission spectrum, as illustrated in Figure S7.

Temperature Measurement and Calibration

Real-time temperature was monitored using either an external type-K thermocouple or the integrated Pt-RTD, depending on the device configuration. For plasmonic structures without an integrated RTD (e.g., cAu-IM or Au thin film), surface temperature was measured using a thermocouple in contact with the top surface. For the assembled thermocycler incorporating a Pt-RTD, the embedded RTD was used for temperature monitoring. Calibration of the Pt-RTD was performed by measuring resistance using a digital multimeter while heating the thermocycler on a hot plate. The detailed calibration process is described in Figure S9.

Plasmonic Photothermal Characterization with Single-Color LEDs

Five single-color LEDs with peak wavelengths at 477 nm (blue), 527 nm (green), 539 nm (lime), 593 nm (PC amber), and 634 nm (red) (LXML-PB01-0040, LXML-PM01-0100, LXML-PX02-0000, LXM2-PL01-0000, LXM2-PD01-0060, Lumileds) were characterized by their emission spectra using a spectrometer (C10082CAH, Hamamatsu). Photothermal temperature enhancement efficiency was evaluated under each single-color LED illumination across the visible spectrum. Detailed methodology and results are provided in Note S1, Figure S6, and Table S2 of the Supporting Information.

Thermocycling Dynamics

Heating and cooling rates during two-step thermocycling (60–95 °C) were calculated from temperature–time profiles. The heating rate was defined as the temperature increase (ΔT = 35 °C) divided by the duration of the heating phase, and the cooling rate was similarly determined from the cooling phase.

PCR Mixture Preparation

The PCR mixtures containing 1 μL of template, 4 μL of 5 × DF reaction buffer (optimized blend of Mg2+, dUTP, and dNTP, DirectFast qRT-PCR kit, NanoHelix Co., Ltd.), 2 μL of DF enzyme mix (optimized blend of reverse transcriptase, antibody-coupled Taq DNA polymerase, RNase inhibitor and heat-labile uracil-DNA-glycosylase, DirectFast qRT-PCR kit, NanoHelix Co., Ltd.), 1 μL of forward primer (10 μM, 2019-nCoV_N1–F, IDT Inc.), 1 μL of reverse primer (10 μM, 2019-nCoV_N1-R, IDT Inc.), 0.5 μL of TaqMan probe (10 μM, 2019-nCoV_N1–P, IDT Inc.), and 2 μL of bovine serum albumin (BSA) (10 μg/μL, Sigma-Aldrich) were brought to a final volume of 20 μL with RNase-free water. For preliminary performance evaluation, a plasmid containing the N gene of SARS-CoV-2 (2019-nCoV_N_Positive Control, IDT Inc.) was used as the template. No-template controls (NTCs) were prepared by replacing the template with RNase-free water.

PCR Thermocycling Conditions

For plasmid control, plasmonic PCR thermocycling was performed without a reverse transcription step, consisting of denaturation at 91 °C for 0 s and annealing/extension at 60 °C for 1 s. For comparison, conventional benchtop PCR using plasmid control was conducted on a StepOne Plus Real-time PCR system (Applied Biosystems, Thermo Fisher Scientific Inc.) with denaturation at 95 °C for 1 s and annealing/extension at 60 °C for 10 s, the minimum programmable duration. For clinical RNA samples, a reverse transcription step was included at 50 °C for 2 min prior to thermocycling, which followed the same plasmonic cycling protocol.

Thermal Uniformity Analysis

Infrared thermal imaging was performed to evaluate the spatial temperature distribution of the 2 μL PCR chamber during plasmonic thermocycling. Thermal images were acquired using an infrared thermal camera (FLIR E75, FLIR Systems, Inc.) under the same denaturation and annealing/extension temperature set points used for plasmonic DRT-qPCR. Temperature variation across the reaction chamber was analyzed directly from the thermal images.

Quantitative PCR Signal Analysis

Fluorescence images were captured through a microlens array camera at each PCR cycle. For real-time amplification analysis, fluorescence intensity was quantified by masking the bubble-free region of the reaction chamber in each image and subtracting the background signal obtained from the first cycle. C_t value was defined as the cycle number where the fluorescence signal crossed a threshold line set at seven times the standard deviation of the baseline (cycles 1–15), provided that the maximum fluorescence intensity exceeded 100. Amplification efficiency (E) was calculated from the slope of the standard curve generated by serial dilutions, using the formula E (%) = (10^–1/slope – 1) × 100.

Clinical Validation of Plasmonic DRT-qPCR

A total of 60 nasopharyngeal swab specimens in viral transport medium (VTM; Copan Diagnostics Inc.) (31 positives and 29 negatives for SARS-CoV-2), precharacterized using the Xpert Xpress SARS-CoV-2 assay (Cepheid), were used for clinical validation. All specimens were deidentified prior to analysis to ensure patient anonymity. To verify sample quality, benchtop RT-qPCR (CFX96, Bio-Rad) was performed to confirm successful amplification of the internal control (RNase P). Only samples with valid RNase P signals were used for N1 gene detection with the plasmonic DRT-qPCR system. Samples were aliquoted and stored at −80 °C until use. All experimental procedures were conducted in a Class II biosafety cabinet within a Biosafety Level 2 (BSL-2) laboratory.

Statistical Analysis

Data described as mean ± standard deviation (SD) represent results from at least three independent experiments.

The experimental data supporting the findings of this study are available in the manuscript and the Supporting Information. Raw data are available from the authors upon reasonable request.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsami.6c04810.

• Comparative overview of representative ultrafast photothermal PCR platforms and commercial POC PCR systems, optical and transmission electron microscopy (TEM) images of the PNT and cAu, FDTD simulations of cAu and nanocavity absorption, measured optical spectra of Au and cAu-IM with varying thicknesses, calculation of WLED absorption efficiency, photothermal performance of cAu-IM versus Au film under single-color LED illumination, schematic of LED-driven temperature feedback control, thermocycling performance as a function of cAu, SiO₂, and Au film thicknesses, calibration and circuitry of integrated Pt-RTD sensors, schematic and photograph of the integrated point-of-care RT-qPCR system, primer and probe sequences, temperature distribution of the 2 μL PCR chamber, amplification performance of benchtop and plasmonic PCR, comparison of extraction-free and conventional RT-qPCR protocols, clinical validation results including C _t values, diagnostic performance, and retest outcomes, representative amplification curves of clinical samples, and operational configuration during plasmonic PCR testing (PDF)

H.J. and K.-H.J. conceived the idea and designed all experiments. H.J. designed and fabricated the PNT. H.J. and E.J. optimized the direct RT-qPCR protocol and conducted the clinical validation. J.-M.K. designed and fabricated the microlens array fluorescence camera. H.J., E.-S.Y., and H.-J.C. produced the disposable PCR cartridges and prepared the PCR mixtures. H.J. and J.P. calculated the optical properties and working principles of the plasmonic nanocavity through numerical modeling. H.J. and H.N. explored thermal behavior of the plasmonic nanocavity using finite element modeling. B.-H.K. and K.H.S. developed the point-of-care RT-qPCR system. D.R. mass-produced the plastic body and adhesive layers for the cartridges. H.J. and K.-H.J. analyzed and wrote the manuscript. K.-H.J., T.Y.K., and M.K. supervised the project. All authors reviewed the manuscript and approved the final version.

Institutional Review Board Statement This study was approved by the Institutional Review Board (IRB) of Samsung Medical Center (IRB No. SMC 2024-06-130).

The authors declare no competing financial interest.