I. Core Research Summary

• Support-free aerosol‑self‑standing 3D gold microcolumn array

No sacrificial layers or support structures are required; a layered stacking approach yields micrometer‑scale gold pillars with high porosity, each pillar approximately 250 μm tall and 73 μm in diameter. This three‑dimensional architecture significantly increases the electrode’s specific surface area, resulting in an electrochemical redox current that is 170% higher than that of a conventional 2D planar electrode.

• Fly‑moles‑level ultra‑high detection sensitivity

Dual-sensor chips were constructed by immobilizing the SARS‑CoV‑2 S1 and RBD antigens, respectively. The detection limits for anti‑S1 antibodies and anti‑RBD antibodies were 2.8 fM and 16.9 fM, enabling the capture of antibodies at extremely low concentrations in human serum.

• Ultra-fast detection in seconds, with no need for prolonged incubation.

After the sample is introduced into the microfluidic channel, a saturation signal exceeding 90% is achieved within 11.5 seconds, eliminating the need for ELISA’s several‑minute incubation step and enabling on‑site, real‑time readout when used with a portable electrochemical smartphone device.

• Reproducible and Regenerable Sensing Interface

pH 2.5 formic acid etching for just 1 minute can dissociate antigen–antibody complexes; after nine regeneration cycles, the sensor signal degrades by less than 5%, enabling multiple assays on a single chip and reducing consumable costs.

• Excellent specificity and batch-to-batch consistency

There is no significant cross‑reaction with interfering substances such as IL‑6 or the nucleoprotein N antibody; the relative signal deviation among six chips from the same batch is only 3%–7%, demonstrating outstanding stability of the digital aerosol printing process.

II. Core Operating Principle of Aerosol Spraying and Its Unique Process Advantages

(1) Complete Fabrication Process of the Sensor Chip

• A Cr/Au conductive underlayer is deposited on the substrate to serve as a three-electrode base.
• Ultrasonic atomization of gold nanoparticle ink, with nitrogen as the carrier gas and a coaxial sheath gas for precise focusing, enables CNC‑controlled, layer‑by‑layer annular printing, yielding self‑supporting 10×10 gold micropillars through solvent‑induced surface tension.
• A 400°C high-temperature sintering process is employed to remove the binder, yielding a porous, highly conductive three-dimensional pillar structure.
• Drop‑coat an rGO dispersion, activate the carboxyl groups with EDC/NHS, covalently couple the SARS‑CoV‑2 S1/RBD antigen, and block non‑specific binding sites with BSA.
• PMMA molds are used to replicate PDMS microfluidic channels, which are then bonded to the electrode substrate to complete chip assembly.

(II) Five Core Processes (in comparison with lithography, wet etching, and screen printing)

• Support-free self-standing 3D stereoscopic electrode direct writing

Conventional photolithography can only fabricate planar structures or shallow trenches, whereas aerosol‑based layer‑by‑layer ring‑printing leverages surface tension to self‑assemble micron‑scale pillars, eliminating the need for molds or sacrificial layers and enabling the one‑step fabrication of high aspect‑ratio three‑dimensional sensing structures.

• Ultra-high specific surface area with controllable shaping

The microcolumn height and array density can be freely tuned, and the porous gold scaffold, combined with rGO‑induced secondary nano‑folds, dramatically increases antigen loading capacity, thereby significantly lowering the detection limit.

• Micrometer-scale digital maskless customization

CAD allows free modification of array layouts and pillar dimensions; a new sensor configuration can be printed in just 30 minutes, whereas photolithographic fabrication typically takes several days, enabling rapid iteration across a wide range of pathogen‑specific chips.

• Mild shaping compatible with subsequent bio‑modifications.

The entire printing and sintering process preserves the conductive layer of the substrate, while the porous structure on the microcolumn surface facilitates rGO adsorption and covalent binding to antigens, resulting in a simple and efficient biofunctionalization step.

• Extremely low material waste, enabling low-cost mass production.

Noble metal gold ink is deposited only in the sensing area, which significantly reduces material consumption compared to a full‑area evaporated gold film, making it well suited for large‑scale mass production of grassroots POCT assay chips.

III. Five Key Technological Innovations

Innovation 1: A Novel Self-Supporting 3D Gold Micropillar Fabrication Process via Aerosol Ring Printing

Innovation 2: Three-dimensional pillar + rGO double-layer nanocomposite sensing interface

Innovative 3-Second Impedance-Based Immunosensing Detection System

Innovation 4: A Recyclable and Reusable Microfluidic Sensing Chip

Innovation: A Fully Digital, Portable Diagnostic Chip Manufacturing Roadmap for Aerosols

IV. Existing Process Challenges and Corresponding Solutions for Overcoming Them

Challenge 1: Traditional microfabrication of three-dimensional electrodes is extremely costly and time-consuming.

Challenge 2: Planar electrodes have a small specific surface area, resulting in a relatively high detection limit for antibodies.

Challenge 3: Conventional microarrays require incubation for several tens of minutes, resulting in low assay throughput.

Challenge 4: Sensor chips are mostly single-use, resulting in high detection costs.

Challenge 5: For lithography‑based chips, any structural modification requires a new mold, resulting in slow iteration.

V. Future Diversified Industrial Application Scenarios

1. Rapid Immunoassay Screening at the Grassroots Level / Port of Entry

2. Quantitative Assessment of Vaccine Immunogenicity

3. Development of a Rapid Diagnostic Chip for Emerging Infectious Diseases

      4. On-site Monitoring of Animal Diseases

5. In vitro micro‑scale screening of tumor markers

Figure 1. Schematic illustration of the fabrication process for a 3D‑printed COVID‑19 detection chip (3DcC) using aerosol jet printing of nanoparticle inks. a) A patterned gold film is deposited on the surface of a glass substrate, serving as the base for the working electrode (WE), counter electrode (CE), and reference electrode (RE) of the 3DcC electrochemical cell; b) Principle of operation of the aerosol jet printer: ultrasonic energy converts gold ink into an aerosol composed of microdroplets, which are conveyed to the nozzle by nitrogen (N₂) and then focused onto the gold film of the working electrode under the action of a sheath gas (also N₂), forming micropillar structures; the entire process is digitally controlled via CAD software; c) An AJ‑printed 10×10 array of gold micropillars, with inter‑pillar spacing indicated; d) AJ‑printing details of a single micropillar: leveraging the surface tension (γ) of the printing ink to achieve rapid, layer‑by‑layer stacking of nanoparticle ink micro‑rings; no supporting structures are required throughout the process; after each layer is printed, solvent is removed by heat from a pressure plate (heated to 150°C using a custom heater), leaving a dried ink layer that serves as the substrate for the next micro‑ring, in a cyclical manner; e) Manufacturing process of the PDMS housing for the 3DcC device: a PDMS structure is fabricated using a PMMA master mold via replica molding, with cross‑sectional dimensions of 1 × 1 × 5 mm³ for section A and 2 × 1 × 10 mm³ for section A1. This structure subsequently functions as a negative mold for the PDMS housing, featuring internal cavities designed to accommodate microfluidic channels (as shown); f) The 3DcC device is assembled by placing the PDMS housing—with its integrated microfluidic channels—alongside the micropillar electrodes on the glass substrate. Prior to this step, the micropillar electrodes have been functionalized with reduced graphene oxide (rGO) and viral antigens, following the procedure described in Figure 2.

Figure 2. Functionalization of the 3D‑printed microelectrode and the operating principle of the 3DcC sensor. a) Au‑microcolumns printed by additive manufacturing prior to surface treatment (Step 1). b) The electrode is coated with carboxylated reduced graphene oxide (rGO) sheets via a simple drop‑casting process (Step 2). SEM images show the modification of the gold columns by the rGO layers; electrostatic or van der Waals interactions link the rGO sheets to the microcolumns, a process likely facilitated by the porous surface structure of the 3D‑printed microcolumns, as illustrated in Figure 3c. The figure also depicts the molecular structure of the rGO layer, highlighting carboxyl and hydroxyl functional groups. c) Viral antigens are conjugated to the rGO sheets using an EDC–NHS chemistry (Step 3): EDC and NHS molecules activate the carboxyl groups on the rGO, enabling amide bond formation between the rGO’s carboxyls and the antigen’s amines, thereby immobilizing and anchoring the SARS‑CoV‑2 recombinant antigen onto the rGO layer; both the spike protein S1 and the RBD antigen are fixed in this manner. Subsequent treatment of the column surface with bovine serum albumin (BSA) blocks non‑specific binding sites on the sensor. d) Antibodies selectively bind to their specific antigens within the sensor through antibody–antigen interactions (Step 4). e, f) Schematic diagrams illustrate the sensing mechanism of the 3DcC device. It is anticipated that, during redox reactions at the electrode/electrolyte interface, an electrical double layer (CDL), the inner Helmholtz plane (IHP), the outer Helmholtz plane (OHP), and a diffusion layer will form. An equivalent circuit diagram is shown. Upon introduction of antibodies (Figure f), the antibodies rapidly bind to the antigens on the electrode surface, altering the Nyquist plot (schematically depicted in Figures e and f); this change can be detected via electrochemical impedance spectroscopy (EIS).

Figure 3. Physical and chemical characterization of the 3DcC device. a) Optical image of the 3DcC device obtained via the fabrication process described in Figures 1 and 2; b) Optical and scanning electron microscopy (SEM) images of an AJ‑printed gold micropillar array (10×10) at different magnifications; c) SEM topographic images of a single gold micropillar at various magnifications: the pillar tip is sharply pointed, with a small circular depression (diameter 20 µm) at the apex, a feature arising from the printing process shown in Figure 1d; the pillar surface consists of gold clusters, forming a characteristic porous structure that facilitates the attachment of rGO flakes to the electrode surface; d) SEM morphology of gold micropillars after rGO modification: rGO layers form a secondary three‑dimensional network around the pillars and exhibit wrinkling due to π–π interactions between graphene sheets; e) Raman spectra of uncoated, rGO‑coated, and antigen‑immobilized gold micropillars on the rGO–Au surface: the spectra of the coated samples display both defect‑related peaks and the graphite peak. Following S1 antigen immobilization, the graphite peak shifts to higher wavenumbers; f) EDX analysis reveals the elemental composition—expressed as atomic percentages—before coating, after rGO encapsulation, and after antigen immobilization on the rGO–Au surface; g) The 3DcC device is connected to a portable potentiostat, which interfaces with a smartphone via a USB‑C port and records signals using the PStouch software.

Figure 4. Detection of SARS-CoV-2 spike protein S1 antigen–antibody complexes at different molar concentrations and under regeneration conditions. a) Nyquist plots obtained using the EIS method for the 3DcC sensor in PBS solution, with and without S1 antigen–antibody complexes at concentrations of 0.01 × 10⁻¹⁵ M, 1 × 10⁻¹⁵ M, 1 × 10⁻¹² M, 100 × 10⁻¹² M, 1 × 10⁻⁹ M, 10 × 10⁻⁹ M, and 30 × 10⁻⁹ M. b, c) Nyquist plots similar to those in (a), acquired after two consecutive regenerations of the sensor using a low-pH chemical solution containing 1.0 M formic acid (pH 2.5); the regeneration process was completed within 60 seconds. For all concentrations, the signals in (b, c) remained within 95% of those in (a). d) In panels (a–c), the charge-transfer resistance (Rct) of the 3DcC sensor before and after each regeneration, measured for each antibody concentration and for control serum. Fetal bovine serum (FBS) and rabbit serum (RS) were used as reference biological fluids in panels (a–d). All measurements were performed in a 50 × 10⁻³ M PBS solution (pH 7.4) containing an equimolar concentration (5 × 10⁻³ M) of ferrocyanide/ferricyanide redox mediator. For each antibody concentration, three consecutive readings were recorded to generate three data sets, with a measurement frequency range of 1 to 10,000 Hz. No incubation time was applied during any of the measurements. The Rct values in (d) were derived by fitting the data from (a–c) to the Randles equivalent circuit shown in Figure 2e.

Figure 5. Detection of SARS-CoV-2 receptor-binding domain (RBD) antigen–antibody complexes at different molar concentrations and under regeneration conditions. a) Nyquist plots obtained by electrochemical impedance spectroscopy (EIS) for the 3DcC sensor in PBS solution, with and without RBD antibodies at concentrations of 1 × 10⁻¹⁵ M, 1 × 10⁻¹⁴ M, 1 × 10⁻¹² M, 100 × 10⁻¹² M, 1 × 10⁻⁹ M, 10 × 10⁻⁹ M, and 20 × 10⁻⁹ M. b, c) Nyquist plots similar to those in (a), acquired after two consecutive regenerations of the sensor using a low-pH chemical reagent consisting of 1.0 M formic acid (pH 2.5); the regeneration process was completed within 60 seconds. In all concentration ranges, the signals in (b, c) remained within 95% of the corresponding signals in (a). d) Charge-transfer resistance (Rct) values for the 3DcC sensor, measured before and after each regeneration step, for each antibody concentration and for control serum, as derived from the data in (a–c). Fetal bovine serum (FBS) and rabbit serum (RS) were used as reference biological fluids in Figures (a–d). All measurements were performed in a 50 × 10⁻³ M PBS solution (pH 7.4) containing an equimolar concentration (5 × 10⁻³ M) of ferrocyanide/ferricyanide redox mediator. Three independent readings were taken at each antibody concentration, with a frequency sweep ranging from 1 to 10,000 Hz. No incubation time was applied during any of these measurements. The Rct values in (d) were calculated by fitting the data from (a–c) to the Randles equivalent circuit shown in Figure 2e.

Figure 6. Investigation of the detection time and regeneration performance of the 3DcC device. a, b) Detection-time profiles for S1 and RBD antibodies obtained with twelve different 3DcC devices during EIS measurements. The figures show how the device impedance evolves over time: the S1 and RBD sensors reach 93.2% and 92% of their saturation impedance values, respectively, within just a few seconds, enabling rapid signal acquisition. In this experiment, the antibody concentration was set to 1 × 10⁻⁹ M, and the data were collected over a frequency range of 1–10,000 Hz. (Zim) Data recording for the device commenced 3 seconds after antibody addition, as it is necessary first to overcome the solution resistance Rs before measuring the charge-transfer resistance (see the schematic in Fig. 2e). c, d) Regeneration‑performance studies reveal the Nyquist plots and changes in charge-transfer resistance for the 3DcC device detecting the S1 antibody: nine regeneration cycles were conducted, with the sensor sequentially incubated for 60 seconds in PBS, in PBS containing 1 × 10⁻⁹ M S1 antibody, and in 1 M formic acid; Figure d illustrates the trend of charge-transfer resistance as a function of the number of regeneration cycles. e, f) Regeneration‑performance data for the 3DcC device detecting the RBD antibody: both sensors exhibited only a slight decline in performance after nine regeneration cycles.

Figure 7. Cross-reactivity and reproducibility studies of the 3DcC device. a) Cross-reactivity testing of the 3DcC device for detecting S1 spike protein antibodies: Nyquist plots were generated for multiple antigens and antibodies, with and without the presence of S1 spike protein antibodies; experiments employed IL‑6 or IL‑6 antigen (0.05 × 10⁻⁹ M), nucleocapsid (N) antibody (1 × 10⁻⁹ M), and RBD antibody (0.1 × 10⁻⁹ M). b) Rct values calculated from the Nyquist plots in panel (a) indicate that the 3DcC device exhibits minimal interference from other related proteins. c) Cross-reactivity testing of the 3DcC device for detecting RBD antibodies: Nyquist plots were constructed for various antigens and antibodies, with and without the presence of RBD antibodies; experiments used IL‑6 antigen (0.05 × 10⁻⁹ M), N‑antibody (1 × 10⁻⁹ M), and spike S1 antibody (0.1 × 10⁻⁹ M). d) Rct values derived from the Nyquist plots in panel (c) likewise demonstrate that the device shows very limited cross‑interference with other relevant proteins. e, f) Under conditions containing S1 antibodies (at a concentration of 1 × 10⁻⁹ M in PBS), six different 3DcC sensors were subjected to reproducibility tests: sensor-to-sensor variability was assessed by calculating the Rct values of each sensor, with a variation range of approximately 7% (error bars represent data from at least three replicate measurements). g, h) Reproducibility test data for six distinct sensors detecting RBD antibodies (at a concentration of 1 × 10⁻⁹ M in PBS); in this case, inter‑sensor variability was approximately 5%.

Figure 8. Kinetic analysis of antigen–antibody interactions. a) Nyquist plots of spike protein S1 antibodies at concentrations of 10 × 10⁻⁹ M (a) and 30 × 10⁻⁹ M on the microcolumn surface, illustrating the association phase, equilibrium phase, and regeneration phase; b) Changes in the Rct values of spike protein S1 antibodies during antigen–antibody binding at concentrations of 10 × 10⁻⁹ M and 30 × 10⁻⁹ M; repeated experiments 1, 2, 3, and 4 correspond to replicate curves obtained during the equilibrium phase; d) Schematic representations of the association, equilibrium, and regeneration phases depicted in panels (a), (b), and (c): During the association phase, the target antibodies (at 10 × 10⁻⁹ M and 30 × 10⁻⁹ M) are loaded into the 3DcC device together with buffer solution, allowing antibody–antigen binding; however, unbound antibody may remain in the solution near the sensor surface, thereby affecting the Rct value. In the equilibrium phase, the 3DcC device is washed with fresh PBS lacking antibodies, which removes unbound antibody from the solution and slightly reduces the Rct value. In the regeneration phase, the sensor surface is regenerated using formic acid solution (1 M; pH 2.5), eluting the antibody from the antigen and resulting in a substantial decrease in the Rct value.