Integrated five-stage cascade impactor with MIP-modified QCM sensor for real-time monitoring of airborne microorganisms Authors: Vahidreza Nafisi, Zahra Mousavian, Alireza Balaei, Yasamin Bide, Ensieh Fahimi Kashani Published: 2026 ABSTRACT Pathogens present a significant threat to global public health, necessitating the development of rapid and reliable detection methodologies for effective monitoring and outbreak management. This study introduces a new sensing device that uses molecularly imprinted polymer (MIP) for sensitive and specific Staphylococcus aureus detection. The research presents a comprehensive investigation into the design and optimization of a five-stage Sioutas Cascade Impactor (SCI) integrated with quartz crystal microbalance (QCM) sensing, enabling continuous, size-resolved collection and in situ monitoring of airborne bacteria. The study combines theoretical modeling of particle impaction dynamics with empirical evaluations of sampler collection efficiency, pressure drop characteristics, and sensor surface properties. A key focus is the optimization of flow control, along with a rigorous evaluation of MIPs as biosensing coatings on QCM sensor. For this purpose, in situ chemical polymerization of dopamine was conducted in the presence of S. aureus bacteria. To confirm surface modification, various microscopic and spectroscopic analyses were employed. Results demonstrate that careful control over flow conditions (14.6 min of sampling time and a flow rate of 6.96 L/min) and resulting pressure gradients, together with MIP surface functionalization, is crucial for achieving high bacterial capture. This work establishes a robust framework for the development of real-time, reusable biosensors for proactive airborne pathogen surveillance. Introduction Airborne pathogenic microorganism’s, such as Pseudomonas aeruginosa, Legionella pneumophila, Staphylococcus aureus, and Bacillus anthracis spores, cause a variety of infectious disorders1. The airborne route is a major mechanism for these diseases to spread because particles that carry them can stay in the air for a long time, allowing them to spread over long distances. Airborne particles can remain suspended for extended periods attributable to their low settling velocities, exemplified by 3.1 × 10–3 m/s for typical 10-μm particles and 3.5 × 10–5 m/s for typical 1-μm particles2,3. Airborne transmission plays a vital role not just in the spread of bacterial diseases but also in viral outbreaks, many of which have developed into global pandemics—such as influenza, which is estimated to cause up to 646,000 deaths globally each year4. The increased potential of bioterrorist attacks employing the aerosol spread of hazardous microorganism’s has further highlighted the necessity for rapid detection and identification of airborne microorganisms5. To reduce the chance of widespread outbreaks, these infections must be identified quickly and precisely. These capabilities enable the prompt implementation of suitable treatments, such as early treatment, effective decontamination, infection control methods, and the deployment of protective countermeasures, while also avoiding personnel exposure and surface pollution. However, the sensitivity and specificity needed for real-time monitoring are frequently omitting in conventional sample approaches like the culture-based methodology. Improved methodologies are consequently necessary to precisely assess the concentration and size distribution of pathogen-laden particles as well as to appreciate the dynamics of aerosolized microorganism transmission in connection to the possible health issues to humans and animals. This information is essential for efficient infection control6. Based on sampling principles, the two primary types of apparatus considered suitable for risk zones with normal (low-level) biocontamination are impact samplers and filtration samplers7. Two cascade impactors, Sioutas Cascade Impactor (SCI) and Andersen Cascade Impactor (ACI), are well-established tools for capturing particles based on their aerodynamic size. The ACI typically employs six to eight impaction stages to separate particles operating at a flow rate of 28.3 L/min. Each stage includes orifice plates that direct airflow onto corresponding collection plates. The SC utilizes four or five impaction stages. Each stage consists of slit plates. It has been optimized for the collection of ultrafine particles and typically operates at 9 L/min. For applications involving ultrafine particle sampling (including microorganisms), environmental pollution monitoring, and personal exposure assessment, the SCI is often preferred due to its efficiency in capturing smaller particles at lower flow rates and its suitability for portable use8,9. On the other hand, Quartz Crystal Microbalance (QCM) sensors have emerged as a promising technology for the detection of airborne pathogens. QCM sensors operate based on the piezoelectric effect, wherein mass changes at the crystal surface induce corresponding shifts in resonance frequency, enabling the detection of bound analytes10. These sensors offer several advantages, including rapid response times and the potential for high-throughput operation10,11. Recent advances are expanding the potential of QCM biosensors for real-time, point-of-care pathogen detection, environmental monitoring and medical diagnostic12. QCM biosensors have been employed to detect a range of airborne bacteria and viruses, including influenza, SARS-CoV-2, and Mycobacterium tuberculosis11,13. The integration of cascade impactors with mass-sensitive devices such as QCM sensors, can be a promising approach for achieving rapid and selective detection of airborne pathogens. Unlike traditional impactor plates, which require post-experimental analysis, QCM sensors enable real-time measurements of mass accumulation, making them especially advantageous for continuous pathogen detection and environmental monitoring. However, several challenges remain. These include optimizing operating flow rates, managing the impact forces and shear stresses experienced by captured bacterial cells, and ensuring that sensor coatings provide strong adhesion of target analytes without compromising cell viability or structural integrity. Precise control of air velocity and pressure drop is essential to minimize disturbances during particle deposition and to ensure efficient capture. These issues have been investigated in this study. Molecularly imprinted polymers have various desirable features that make them attractive for biosensor applications. These include superior thermal and chemical stability, a long shelf life, fabrication repeatability, mechanical durability, reusability, high sensitivity, and outstanding target selectivity. Because of these qualities, MIPs are desirable candidates to be incorporated into biosensing platforms that are intended to identify certain biological targets. In the context of bacterial detection, MIPs function through shape/size-selective identification, which is particularly beneficial for identifying various bacterial species. The basis for this recognition mechanism is the production of template-shaped cavities in the polymer matrix that are exactly complementary to the target bacteria in terms of size, shape, and functional group arrangement. As a result, MIPs can effectively differentiate between bacterial morphologies. For instance, they can recognize spherical S. aureus cells and rod-shaped Escherichia coli cells through physical and chemical matching. Moreover, the rigidity of bacterial cell walls supports the selective interaction with these imprinted cavities, especially under dry or semi dry conditions and optimized air flow. This makes MIP-coated surfaces especially advantageous in air-sampling biosensors or environmental sensors where traditional receptor-ligand interactions may be limited due to fluctuating humidity or surface biofouling. Tokonami et al. used the fundamental differences in surface chemistry between Gram-positive and Gram-negative bacteria to create highly selective detection systems that could distinguish Escherichia coli and Pseudomonas aeruginosa from Bacillus subtilis and S. aureus. In their work, specific emphasis was placed on retaining the electrical conductivity stability of conducting polymers during the electro-polymerization of anionic bacterial strains, a modification that considerably enhanced the efficacy of surface imprinting techniques14. Agar et al. introduced a dual layer sensor platform that combines aptamers with MIPs for the selective detection of S. aureus and E. coli. Their study demonstrated high target specificity for both sensor types, owing to the high affinity of aptamers for their corresponding bacteria and the tailored cavity structure of the MIPs. Aptamers successfully bonded to their respective targets in bacterial mixtures, whereas the imprinted cavities in MIPs allowed for selective bacterial recognition. This hybrid technique, incorporating biological (aptamer) and synthetic (MIP) recognition elements, considerably reduced nonspecific binding while improving the accuracy and reliability of bacterial detection15. Recently, a combination of QCM sensor with MIP coating has been used for the detection of Galectin-3 in serum16. The advantage of this method is the increased speed of detection and the lack of need for labeling the target substance. In another approach, to detect the residual amount of the antibiotic levofloxacin in blood, an electrochemical sensor composed of TiO2 and MIP has been designed and manufactured17. The results indicate that this biosensor not only has high sensitivity but also possesses reproducibility and long-term stability. Other studies have also utilized MIP coating on their studied biosensors to improve sensitivity and stability18–20. In all of these studies, the medium for the presence of the target material/bacteria and its identification is a liquid medium (blood, serum, etc.), and there have been few studies on the detection of airborne pathogens. In this research, an attempt was made to design and construct a device for detecting airborne pathogens by combining the characteristics of a cascade impactor and QCM sensors coated with MIP, and the research results indicate that this has been achieved. Naturally, improving the coating manufacturing process could lead to increased efficiency of this system and improve its sensitivity and specificity. Material and methods Sampler module The sampler contains of a customized five-stage cascade impactor, a vacuum pump, and an array of QCM sensors (at stages B through E, there is no QCM sensor at stage A), illustrated in Fig. 1. Each stage was fabricated from stainless steel 304 using the Electrical Discharge Machining (EDM) process (Rivand 80A, made in Iran), in which the spark head-electrode was made of copper and machined by CNC (Fig. 2). The setting of the EDM machine were: 1 Amp, Frequency L5, Arc control L3).Fig. 1The cascade impactor and nozzle parameters of each stage of the impactor. W = width (mm), L = length (mm), T = depth of the nozzle (mm), S = distance between the nozzle and the impaction surface (mm).Fig. 2Fabricated stages from stainless steel 304 using spark method. Left to right: Stage A to E. The cascade impactor and nozzle parameters of each stage of the impactor. W = width (mm), L = length (mm), T = depth of the nozzle (mm), S = distance between the nozzle and the impaction surface (mm). Fabricated stages from stainless steel 304 using spark method. Left to right: Stage A to E. The hydraulic diameter (Dh) of each nozzle is calculated as:1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${D}_{h}= \frac{4A}{P}\,\, with\,\, P=2\left(W+L\right) \,\,and \,\, A=W\times L$$\end{document}where W and L are the width and length of the nozzle exit. Also the D50 (particle cut-off diameter) for each stage, as a function of hydraulic diameter, is given by21:2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${D}_{50} \cong 6.07\sqrt{\frac{{D}_{h}}{v\times {f}_{c}}} (\upmu \mathrm{m})$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${D}_{h}$$\end{document} is hydraulic diameter of the nozzle (mm), v is exit velocity (m/s) and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${f}_{c}$$\end{document} is Cunningham slip correction factor. Table 1 shows the D50 for a highly concentrated microorganism-air mixture. This assumption is valid for our experimental setup and there is a high concentration of microorganisms in the inlet of the cascade impactor. D50 has been calculated for flow of 7 L/min.Table 1Nozzle geometry for each stage.StageW (mm)L (mm)T (mm)S (mm)Dh (mm)D50 (µm)A0.9190.81.01.723.044B0.5210.81.00.981.800C0.36190.81.00.711.237D0.14250.81.00.280.556E0.09260.81.00.180.364 Table 1Nozzle geometry for each stage.StageW (mm)L (mm)T (mm)S (mm)Dh (mm)D50 (µm)A0.9190.81.01.723.044B0.5210.81.00.981.800C0.36190.81.00.711.237D0.14250.81.00.280.556E0.09260.81.00.180.364 Nozzle geometry for each stage. Gold‑coated QCM crystals SC-101 with a fundamental frequency of 6 MHz were supplied by MAXTEK (USA). The resonant frequency of QCM sensor proportionally changes by particle accumulation on its surfaces:3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta f= -C\Delta m$$\end{document}where C is the Sauerbrey sensitivity constant, which for a 6 MHz, 14 mm QCM sensor (with a gold substrate and AT-cut crystal configuration) is 52.9 × 106 Hz/g. Material for sensor coating In the present study, the bio-coatings applied to the QCM sensors are Molecularly Imprinted Polymers (MIPs). Our MIP coating is of the surface-imprinted polymer (SIP) type22. These were engineered to selectively bind target bacteria (S. aureus). The selectivity of the sensor stem from shape-specific surface cavities formed by bacterial imprinting in Polydopamine (PDA). The QCM sensors detect mass loading of captured S. aureus cells on their SIP-coatings. The measured frequency shift of the sensors is proportional to total mass of S. aureus adhering to the MIP layer. It is a label free detection, without antibody interactions or biochemical recognition. The S. aureus strain (PTCC No. 1112) and E. coli (PTCC No. 1399) were supplied by Persian Type Culture Collection (PTCC) of the Iranian Research Organization for Science and Technology (Iran). E. coli was used for evaluation of specificity of the device. For preparation of the mentioned bacterial suspensions, an aliquot of bacteria frozen glycerol stock was first plated on Nutrient Agar (NA) and incubated at 37°C for 18 h. After incubation, a single colony was resuspended in 10 mL of Nutrient Broth and placed in a shaking incubator at 37°C to prepare an overnight culture. One milliliter of the overnight bacterial culture was centrifuged at 13,000 rpm (11,337 × g) for 5 min, and the bacterial pellet was washed twice with PBS (pH: 7.4). The optical density (OD600nm) of the bacterial suspension was measured and adjusted to OD600nm = 0.25 using PBS. Plate counts confirmed that this corresponded to approximately 1 × 10⁸ to 2 × 10⁸ CFU/mL. All reagents were of analytical or synthetic grade. Hydrogen peroxide (30% w/w, Merck) and sulfuric acid (95–98%, Merck) were employed to prepare the piranha solution. Dopamine hydrochloride (≥ 98%) was obtained from Merck. Tris(hydroxymethyl)aminomethane, sodium hydroxide, and hydrochloric acid (analytical grade) were purchased from Merck for preparation of the 20 mM Tris‑HCl buffer (pH: 8.5). Ultrapure deionized water was produced using a Millipore Milli‑Q® system (18.2 MΩ·cm), Acetic acid (analytical grade, Merck) was used for the bacterial removal step. All reagents were selected to minimize contamination or side‑reactions that could affect the dopamine polymerization kinetics and the uniformity of the polydopamine layer. Tris buffer (20 mM, pH: 8.5) was prepared in‑house by dissolving Tris(hydroxymethyl)aminomethane (Tris) in deionized water and adjusting the pH to 8.5 using 1 M NaOH and 1 M HCl solutions. Both pH adjustments were made gradually under constant stirring to ensure accurate buffering. Coating methods There are various methods for creating MIP-coatings tailored to target bacteria. By considering some aspects (such as the quality, cost and ease of implementation) into account, the following procedure was adopted for this project: Before use, the ozone-cleaned QCM sensor was immersed in piranha solution (H₂O₂/H₂SO₄ = 1:3) for 5 min, then thoroughly rinsed with deionized water and dried with nitrogen gas (N₂). A solution containing S. aureus bacteria (0.1 mg/mL) and dopamine was thoroughly mixed in Tris–HCl buffer (20 mM, pH: 8.5). Two milliliters of this mixture was poured onto the QCM chip placed in a vessel. The setup was then maintained horizontally at room temperature for specific durations (6 and 24 h). Afterwards, the QCM coated with polydopamine was washed twice with 5% acetic acid solution for 5 min each time, followed by rinsing with deionized water to remove the bacterial template, and then dried. For comparison, two other coatings were applied, the porcine gastric mucin (CAS #84082-64-4, Merck, DE) and the glycan Asialoganglioside GM2 (GA2, CAS #35960-33-9, BioSynth): Purified porcine gastric mucin was dissolved at a final concentration of 10 mg/mL and homogenized by magnetic stirring for 1 h in order to conduct adhesion studies. 100 µL of this mixture were pipetted onto the sensors and incubated at room temperature for overnight. Some sensors were coated with 100 μL of GA2 at a final concentration of 1 mg/ml, which was then incubated at room temperature for one hour on a shaker set at 550 rpm23. Also, for recycling of the used sensor, the sensor was washed with 5% acetic acid solution for 5 min. This solution removes the captured bacteria without any damage to the MIP coating. The FTIR analysis showed that the MIP coating remain intact after washing by acetic acid solution (refer to Fig. 5 in Result section). Therefore, the sensors are reusable after recycling cycle. Experimental setup The experimental setup used to evaluate the physical efficiency of the sampler is shown in Fig. 3. The system inlet was connected via a drying column to the outlet of a particle generator, which produced monodisperse polystyrene particles with a controlled diameter of 0.3 µm and a density of 1.05 g/cm3. For count reference, a commercial particle counter was used (Lighthouse Handheld 3016).Fig. 3Sampler physical efficiency test bed. Sampler physical efficiency test bed. For measuring biological collection efficiency, the setup depicted in Fig. 4 was employed. This setup is based on the “Aerosol chamber method” described in the BS EN 17141:2020 standard (EN 17141:2020). A bacterial suspension was introduced into a nebulizer, and then the microorganisms were aerosolized into the glass chamber. The volume of nebulizing chamber is smaller than 2 L and nebulizing flow rate is more than 2 L/min. Airborne microorganisms were suctioned into two cascade impactors (one serving as a control and the other as the test sampler) by vacuum pump operation. then drawn through. Fresh air entered the chamber through zero filters to maintain clean airflow.Fig. 4Microbial test bed. Microbial test bed. Evaluation methods and tools Throughout this study, attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy and field emission scanning electron microscopy (FE-SEM) were employed to characterize the functional groups of the modified QCM sensors and to confirm surface modification. ATR-FTIR spectra were recorded using a PerkinElmer spectrometer (USA) over the wavenumber range of 650 to 4000 cm–1. The morphology of the modified QCM sensors was examined using a FE-SEM (MIRA\LMU, TESCAN, Czech Republic). Also, for evaluation of surface characteristics, roughness tester TR200 (Salu Tron GmbH, Germany) and AFM (ARA-AFM, ARA research co., Iran) was used. Additionally, colony counting and fluorescence microscopy were utilized to quantify bacterial adsorption on the sensor surfaces. The binding potential of S. aureus and E. coli surface receptors to the GA2 glycan was first evaluated using molecular docking simulations. In a series of preliminary studies, we also assessed MIP's sensitivity to the target bacteria (S. aureus) in the presence of E. coli at the same concentrations. After determining the sensitivity of MIP and chemical ligands (GA2 and mucin) for the target bacteria, statistical analysis was utilized to assess the effectiveness of the receptors under investigation in trapping S. aureus. Design Expert software version 12 and analysis of variance (ANOVA) as powerful tools to examine relationships between variables were used to analyze and compare several model evaluation methods using data collected through experiments. Results Characterization of QCM coated with polydopamine MIP. To characterize the functional groups and confirm the successful coating of the QCM with polydopamine MIP, ATR-FTIR analysis was performed (Fig. 5). The presence of broad OH and NH stretching peaks at 3300–3500 cm–1, strong C–H stretching vibrations at 2830–2920 cm–1, as well as aromatic C = C and C–N stretching vibrations at 1610–1500 cm–1 identify the indoline/quinone structure typical of polydopamine networks. The C–O stretching vibration between 1280–1320 cm–1 reflects catechol oxidation, while the weak bands around 750–800 cm–1 correspond to C–H out‑of‑plane bending of aromatic rings. Fig. 5ATR-FTIR spectrum of QCM sample coated with polydopamine MIP. ATR-FTIR spectrum of QCM sample coated with polydopamine MIP. FE-SEM analysis was employed to investigate the surface morphology of the PDA-MIP coated sensor (Fig. 6). The images reveal significant surface changes and the presence of porosity following the coating process. The Fig. 6A correspond to the pristine gold QCM surface at magnifications of × 20,000 and × 2000 (scale bars 2 µm and 20 µm, respectively). The Fig. 6B represent the QCM after polymerization and template removal at identical magnifications. Prior to PDA-MIP deposition, the QCM sensor surface appeared smooth and relatively featureless at both low and high magnifications. After modification with the PDA layer, the sensor surface exhibited increased roughness and a more uniform, granular texture, indicating the successful formation of the polymeric film. Before template elution, SEM images display changes, which may be due to bacterial aggregation or biofilm formation on the sensor surface (Fig. 6C).Fig. 6FESEM images of raw QCM (A) and QCM coated with polydopamine MIP (B) with × 2k and × 20k magnifications, and QCM coated with polydopamine MIP before template elution (C). FESEM images of raw QCM (A) and QCM coated with polydopamine MIP (B) with × 2k and × 20k magnifications, and QCM coated with polydopamine MIP before template elution (C). AFM images (5 × 5 µm) of the polydopamine MIP films before (Fig. 7a) and after (Fig. 7b) template removal (washed with 5% acetic acid) show that the average surface roughness increased after removal. Concurrently, QCM thickness decreased by ≈ 10–13 nm, confirming successful extraction of S. aureus templates and formation of bacteria-imprinted cavities within the PDA network. The observed increase in surface roughness and decrease in thickness provide direct evidence of cavity formation within the PDA-MIP layer following the removal of S. aureus templates from the QCM surface.Fig. 7AFM images of QCM coated with polydopamine MIP before (A) and after (B) template removal, 5 × 5 µm (range 5µm, speed 5 lines/s, resolution 1024). AFM images of QCM coated with polydopamine MIP before (A) and after (B) template removal, 5 × 5 µm (range 5µm, speed 5 lines/s, resolution 1024). Sampler physical efficiency results The average equivalent diameters of microbe-carrying particles (MCPs) responsible for colony forming units (CFUs) are typically greater than 1 µm, and a D50 value below 2 µm is considered suitable for effective microbial sampling24. The collection efficiency of developed device is influenced by airflow rate. Experimental tests demonstrated that particles > 1µm, the device achieves more than 98% removal efficiency across the entire tested airflow range of 4.5 to 9 L/min (Fig. 8). For smaller particles, the collection efficiency is more sensitive to flow rate variations and there is an optimal airflow rate (about 7 L/min).Fig. 8Overall device capture efficiency at different flows. Overall device capture efficiency at different flows. In order to investigate the contribution of each stage of the sampler, initially only stage A was placed in the impactor and then subsequent stages were added to achieve the complete five-stage configuration. The result of this experiment is given in Table 2. The results clearly show that the five-stage sampler is capable of capturing a significant fraction of particles as small as 0.3 μm. The smaller particles are predominantly capture in the lower stages with narrower slit geometries designed to target smaller particle sizes through increased inertial impaction efficiency.Table 2Particle capture efficiency (%): role of each sampler stage (at 7 L/min).Level0.3 µm0.5 µmA28.873.6A + B42.879.0A + B + C54.488.8A + B + C + D70.393.0A + B + C + D + E78.296.0 Table 2Particle capture efficiency (%): role of each sampler stage (at 7 L/min).Level0.3 µm0.5 µmA28.873.6A + B42.879.0A + B + C54.488.8A + B + C + D70.393.0A + B + C + D + E78.296.0 Particle capture efficiency (%): role of each sampler stage (at 7 L/min). It is worth mentioning that non-biological particles typically retain their rigid geometry during adhesion to the surface. But impaction force result in the bacterial cell deformation as a function of time t25. The deformation Z(t) is modeled by:4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Z\left(t\right)= \frac{{F}_{imp}}{{k}_{1}}+ \frac{{F}_{imp}}{{k}_{2}} \left[1-\mathrm{e}\mathrm{x}\mathrm{p}(-t\frac{{k}_{2}}{{\eta}_{2}})\right]$$\end{document}where Fimp is the impact force, k1 and k2 are constants related to the cell membrane elasticity (k1 = 0.1 N/m and k2 = 1.2 N/m for Gram-positive bacteria, k1 = 0.026 N/m and k2 = 0.33 N/m for Gram-negative bacteria) and η2 is the viscosity of the cell surface (η2 = 3.0 N.s/m for Gram-positive bacteria and η2 = 0.3 N.s/m for Gram-negative bacteria). The flow within the cascade impactor is approximately laminar, then the impact force (Fimp) and kinetic energy (Ek) of the particle upon impaction are expressed as:5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${F}_{imp}= \frac{{mv}^{2}}{2S}= \frac{m}{2S}{(\frac{Q}{WL})}^{2}$$\end{document}6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{k}= \frac{1}{2}m{v}^{2}$$\end{document}where S is deceleration gap distance (m), Q is volumetric flow rate (m3/s), (W,L) are dimensions of the impactor nozzle (m). Table 3 presents the calculated impact parameters at a flow rate of 7 L/min for a 1-μm diameter particle.Table 3Impact parameters for a 1-µm particle (at 7 L/min).StageFimp (N) × 10–10Ek (J) × 10–12A16.81.7B44.44.4C103.810.4D396.639.7E887.488.7 Table 3Impact parameters for a 1-µm particle (at 7 L/min).StageFimp (N) × 10–10Ek (J) × 10–12A16.81.7B44.44.4C103.810.4D396.639.7E887.488.7 Impact parameters for a 1-µm particle (at 7 L/min). The Eq. 4 reveals that bacteria cells experience progressively deformation when they hit the surface. Table 4 indicates the deformation in the steady state. Based on this theoretical calculation, the particle deformation for a 1-µm particle is not applicable at stage E. Also, this statement is true for gram-negative bacteria at stage D. In result, it seems that bacteria are raptured at these stages. In other situations, this cell deformation could help to better attachment to the surface, because of increasing of contact area.Table 4Particle deformation hitting to the stage surface (at 7 L/min) in the steady state (t → ∞).StageZ(t), (nm)Gram-positive (e.g., S. aureus)Gram-negative (e.g., E. coli)A1870B48184C112431D4301646E9613680 Table 4Particle deformation hitting to the stage surface (at 7 L/min) in the steady state (t → ∞).StageZ(t), (nm)Gram-positive (e.g., S. aureus)Gram-negative (e.g., E. coli)A1870B48184C112431D4301646E9613680 Particle deformation hitting to the stage surface (at 7 L/min) in the steady state (t → ∞). The bio-sensor function Based on our sensor driving circuit the initial frequency for new/uncoated QCM sensor is 5,991,444 ± 188 Hz. After coating the initial frequency is shifted to 5,989,879 ± 100 Hz. These measurements are very dependent on each sensor/coating characteristics. There are some variations but it is not important to our results because of the mass loading is proportional to the frequency shift and the variations in the initial frequency can be adjusted on the calibration phase. The accuracy and long-term stability of the measurement circuit of the QCM crystal are ± 1.7 Hz and ± 7.0 Hz, respectively. Based on Eq. 3, the sensitivity of bare QCM sensor is ~ 30 ng. On the other hand, m is mass of particle which for a spherical particle is given by:7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$m= \frac{\pi }{6}{\rho}_{p}{({D}_{p})}^{3}$$\end{document}where, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\rho}_{p}$$\end{document} is the particle density (kg/m3) and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${D}_{p}$$\end{document}​ is the particle diameter (m). Then, the mass of a bacteria such as S. aureus (Dp ≈1 µm and ρp ≈1100 kg/m3) is about 0.58 picogram. In result, the sensitivity of the bare QCM sensor for bacteria detection is 51,000 bacteria. The device has ~ 100% capture efficiency at 7 L/min air flow rate, then it is equal to a LOD of ~ 7.3 CFU/mL/min airborne bacteria detection. Comparison of two curves in Fig. 9 (stage C and D, no coating) reveals that the stage D is more effective than stage C to capture the bacteria in the same size of S. aureus. It is based on the size-selective feature of the cascade impactor. In addition, proper coating helps to increase capture performance in stage D (curve stage D, MIP coating).Fig. 9The impact of the stage on the bacteria capture performance. The impact of the stage on the bacteria capture performance. The effects of environmental and surface coating conditions on the sensor performance were investigated. The response of two coated/uncoated sensors at stages C from two identical cascade samplers were compared. In Fig. 10a, the sensors were tested under dry airflow condition. In contrast, Fig. 10b shows the sensors response when the airflow was wet. The presence of moisture significantly influences the sensor response, leading to larger signal shifts. In addition, uncoated and MIP-coated sensors were compared. The results demonstrate that MIP-coated sensor exhibited greater stability and reduced sensitivity to wet condition. It means that the coating effectively minimizes the effect of environmental moisture on the sensor performance.Fig. 10The effect of coating on the sensor response. (a) Dry conditions, (b) wet conditions. The effect of coating on the sensor response. (a) Dry conditions, (b) wet conditions. Figure 11 illustrates that surface characteristics (i.e. surface roughness) affect the sensor performance in three independent experiments. The QCM frequency responses are displayed on the left side of the figure, and the microstructural variations of the sensor surfaces are shown in the SEM images on the right. Among the investigated sensors, sensor in third experiment displayed the most significant response, which correlates with its MIP-coated surface with proper roughness. More nanopores improve the sensor's sensitivity and measurement accuracy by increasing the effective surface area and facilitating enhanced interaction with the target microorganism. The findings emphasize that careful optimization of surface porosity and shape is critical for enhancing the performance of biosensors.Fig. 11SEM of the sensor surface after microbial tests. Larger changes in sensor response are associated with sensors that have proper coating. SEM settings for all left images are: HV: 20 kV, MAG: 150 kx, view field: 1,445 nm, WD: 4.983 mm. SEM of the sensor surface after microbial tests. Larger changes in sensor response are associated with sensors that have proper coating. SEM settings for all left images are: HV: 20 kV, MAG: 150 kx, view field: 1,445 nm, WD: 4.983 mm. Statistical analysis for bacterial experiment Using the Historical Data Design (HDD) approach, the Response Surface Methodology (RSM) was used to assess which specific experimental factors affect the biosensor's ability to identify S. aureus. In design of biosensor surface, two important performance requirements must be optimized:S. aureus must bind preferentially and with high affinity to the sensitive detecting area.S. aureus must be efficiently transported from the bulk phase to the sensor surface. S. aureus must bind preferentially and with high affinity to the sensitive detecting area. S. aureus must bind preferentially and with high affinity to the sensitive detecting area. S. aureus must be efficiently transported from the bulk phase to the sensor surface. S. aureus must be efficiently transported from the bulk phase to the sensor surface. In this study, five independent variables were included (Table 5). A total of 50 experimental combinations were analyzed, and the actual values and results are summarized in Table 6.Table 5The factors examined in the historical data design to identify the conditions for capturing bacteria.FactorNameTypeMinMaxCoded lowCoded highMeanStd. devAFlow rateNumeric1.708.00-1 ↔ 1.70 + 1 ↔ 7.003.591.41BSampling time*Numeric0.900015.00-1 ↔ 1.25 + 1 ↔ 15.004.683.14CLayersCategoricCDLevels:2DCoatingCategoricMIPGA2MucinNoneLevels:4ECascade type based on its pressure drops**CategoricLowHighLevels:2Duration of pathogen exposure*, hydrodynamic resistance across stages**.Table 6Historical data design for the influences of the five independent variables on the S. aureus capturing efficiency in experimental results.RunFactor AFactor BFactor CFactor DFactor EResponse: S. aureus146CNoneHigh8410246DNoneHigh935034.56DNoneLow6360473DGA2High193054.56DNoneLow725066.43CNoneLow16,00074.43.5DNoneHigh795085.253.5DNoneLow7000933.5DNoneLow11,0001033CGA2High851133CGA2High1971233DMIPHigh3820133.423DNoneLow10,6501433CMIPHigh2101533DGA2High935165.13.4DNoneLow9600173.53.4CGA2High183.43.4CNoneLow8350193.43.4DNoneLow11,100203.53.4CGA2High100213.53DGA2High384223.41.25DMIPLow1770233.20.9CNoneHigh3620243.650.9DMIPLow6000255.13.3DMIPLow1840261.73.3DMIPHigh580271.73.3DMIPHigh92281.75CMIPHigh1170291.75DMIPHigh790301.95CNoneLow490311.712CMIPHigh2100321.712DMIPHigh12,630331.912DNoneLow22,880341.712.3DMIPHigh1280351.912.3DNoneLow5490361.77.4CMIPHigh1153371.77.4DMIPHigh1711381.97.4CNoneLow3440392.515CNoneHigh4020402.515DNoneHigh4550412.59DMucinHigh693422.69CMucinLow66431.74DMIPHigh150441.78DMucinLow140451.77DNoneHigh8134653CNoneHigh29,0004753DNoneHigh19,550484.53CNoneHigh13,000494.53DNoneHigh11,600504.53CNoneHigh13,200 Table 5The factors examined in the historical data design to identify the conditions for capturing bacteria.FactorNameTypeMinMaxCoded lowCoded highMeanStd. devAFlow rateNumeric1.708.00-1 ↔ 1.70 + 1 ↔ 7.003.591.41BSampling time*Numeric0.900015.00-1 ↔ 1.25 + 1 ↔ 15.004.683.14CLayersCategoricCDLevels:2DCoatingCategoricMIPGA2MucinNoneLevels:4ECascade type based on its pressure drops**CategoricLowHighLevels:2Duration of pathogen exposure*, hydrodynamic resistance across stages**. The factors examined in the historical data design to identify the conditions for capturing bacteria. Duration of pathogen exposure*, hydrodynamic resistance across stages**. Table 6Historical data design for the influences of the five independent variables on the S. aureus capturing efficiency in experimental results.RunFactor AFactor BFactor CFactor DFactor EResponse: S. aureus146CNoneHigh8410246DNoneHigh935034.56DNoneLow6360473DGA2High193054.56DNoneLow725066.43CNoneLow16,00074.43.5DNoneHigh795085.253.5DNoneLow7000933.5DNoneLow11,0001033CGA2High851133CGA2High1971233DMIPHigh3820133.423DNoneLow10,6501433CMIPHigh2101533DGA2High935165.13.4DNoneLow9600173.53.4CGA2High183.43.4CNoneLow8350193.43.4DNoneLow11,100203.53.4CGA2High100213.53DGA2High384223.41.25DMIPLow1770233.20.9CNoneHigh3620243.650.9DMIPLow6000255.13.3DMIPLow1840261.73.3DMIPHigh580271.73.3DMIPHigh92281.75CMIPHigh1170291.75DMIPHigh790301.95CNoneLow490311.712CMIPHigh2100321.712DMIPHigh12,630331.912DNoneLow22,880341.712.3DMIPHigh1280351.912.3DNoneLow5490361.77.4CMIPHigh1153371.77.4DMIPHigh1711381.97.4CNoneLow3440392.515CNoneHigh4020402.515DNoneHigh4550412.59DMucinHigh693422.69CMucinLow66431.74DMIPHigh150441.78DMucinLow140451.77DNoneHigh8134653CNoneHigh29,0004753DNoneHigh19,550484.53CNoneHigh13,000494.53DNoneHigh11,600504.53CNoneHigh13,200 Historical data design for the influences of the five independent variables on the S. aureus capturing efficiency in experimental results. Mucin (Mucin glycoprotein matrix) and GA2 (Galactose-specific lectin) are two bio-coating that was also tested as an alternative coating to MIP. The primary focus of this report is on the MIP-coated biosensor, with further discussion of GA2 to be presented in future work. The relevance and predictive ability of the model created for S. aureus capture efficiency were assessed using regression analysis and Analysis of Variance (ANOVA). Table 7 presents the findings.Table 7Analysis of variance for HDD 2FI model.SourceSum of squaresdfMean squareF-valuep-valueModel19.2272.7516.24 < 0.0001*A—flow rate2.9012.9017.180.0002*B – time0.788810.78884.670.0367*C—stages0.390610.39062.310.1362D—coating10.4733.4920.64 < 0.0001*E—cascade0.006610.00660.03880.8447Residual6.93410.1691Lack of fit6.54370.17691.820.2987Pure error0.388040.0970Cor total26.1548Fit statisticsStd. dev0.4112R20.7349Adeq precision12.79Mean3.36Adjusted R20.6897PRESS10.04C.V. %12.26Predicted R20.6160BIC74.36*Significant result (p-values less than 0.05). Table 7Analysis of variance for HDD 2FI model.SourceSum of squaresdfMean squareF-valuep-valueModel19.2272.7516.24 < 0.0001*A—flow rate2.9012.9017.180.0002*B – time0.788810.78884.670.0367*C—stages0.390610.39062.310.1362D—coating10.4733.4920.64 < 0.0001*E—cascade0.006610.00660.03880.8447Residual6.93410.1691Lack of fit6.54370.17691.820.2987Pure error0.388040.0970Cor total26.1548Fit statisticsStd. dev0.4112R20.7349Adeq precision12.79Mean3.36Adjusted R20.6897PRESS10.04C.V. %12.26Predicted R20.6160BIC74.36*Significant result (p-values less than 0.05). Analysis of variance for HDD 2FI model. *Significant result (p-values less than 0.05). The Model F-value of 16.24 implies the model is significant. There is only a 0.01% chance that an F-value this large could occur due to noise. Sum of squares is Type II Classical. P-values less than 0.05 indicate model terms are significant. In this case A, B, C, D, AB, AD, AE, CD are significant model terms. The F-value for the Lack of Fit is 1.82 that implies the Lack of Fit is not significant relative to the pure error. There is a 29.87% chance that a Lack of Fit F-value this large could occur due to noise. It shows that the model matches the experimental data satisfactorily. In addition, the ANOVA regression model demonstrated a determination coefficient (R2) of 0.73 that explains approximately 73% of the variability in response variables can be explained by this model. The value of the adjusted determination coefficient (R2adj) was 0.68 which showed the strong robustness of the model. Moreover, a low value of coefficient of variation (CV = 12.26%) indicated the reliability and precision of the experiments. A suitable signal to noise ratio was further confirmed by the model's sufficient precision value of 12.79, which was significantly higher than the acceptable threshold of 4.0. The presented logarithmic regression model (Table 8) describes the relationship between the concentration of S. aureus and a series of levels of each factor (Factor A to Factor E) that characterize the biosensor system used for bacterial capture. The model coefficients reflect the relative importance and directionality of each factor’s influence on the bacterial trapping efficiency.Table 8The logarithmic regression model.TermCoefficient estimatedfStandard error95% CI low95% CI highIntercept3.3210.14133.043.61A-flow rate0.645110.15570.33080.9595B-time0.301410.13950.01960.5832C-layers0.095710.0630-0.03150.2229D[1]0.284610.12100.04030.5289D[2]-0.478910.1619-0.8059-0.1520D[3]-0.598710.2005-1.00-0.1937E-Cascade-0.013710.0694-0.15390.1266 Table 8The logarithmic regression model.TermCoefficient estimatedfStandard error95% CI low95% CI highIntercept3.3210.14133.043.61A-flow rate0.645110.15570.33080.9595B-time0.301410.13950.01960.5832C-layers0.095710.0630-0.03150.2229D[1]0.284610.12100.04030.5289D[2]-0.478910.1619-0.8059-0.1520D[3]-0.598710.2005-1.00-0.1937E-Cascade-0.013710.0694-0.15390.1266 The logarithmic regression model. In compatibility with cascade impactor theory and our experimental results (Fig. 8), selection of an optimum flow enhances capture performance which concentrate S. aureus in the coated zones. The model’s coefficients for the coating variables (+ 1.85 for MIP, -0.47 for GA2/Mucin and -0.59 for None) demonstrate the superiority of MIP coatings. Based on Table 7, Factor C (cascade’s stage) and Factor E (cascade’s pressure drops) are statistically insignificant parameters. Also, the model’s coefficients for these parameters are very small (+ 0.09 and -0.01). These mean that bacteria are captured effectively throughout hydraulic gradients and both stage C and D. The interaction between S. aureus and MIP recognition sites is time-dependent. Due to the large number of accessible recognition sites, the binding rate is high at the beginning. The rate of bacterial attachment gradually lowers and eventually stabilizes at an equilibrium state as the number of accessible binding sites decreases. The flow rate in MIP biosensors is essential for delivering S. aureus cells to the sensor surface. The MIPs, with their target-specific cavities, enabled the selective entry of S. aureus, enhancing the sensor's selectivity. While increased flow rates shorten the testing time and facilitate the removal of non-specifically bound cells, lower flow rates prolong the interaction between target bacteria and recognition sites, resulting in improved binding efficiency. Consequently, an intermediate flow rate is optimal for maximizing sensor performance. A flow rate of 6.96 L/min was identified as optimal under these conditions. The RSM analysis demonstrated that Using MIPs as a coating, particularly in the stage D, improved both selectivity and sensitivity in extracting S. aureus from the sample (Fig. 12).Fig. 12Numerical optimization result of parameters on bacterial capturing in biosensor. Numerical optimization result of parameters on bacterial capturing in biosensor. Based on Fig. 8, flow rate of 7 L/min is an optimal air flow to maximize the physical performance. Reduced capture efficiency may arise from higher flow rates in biosensor systems because they may reduce the amount of time target analytes (like S. aureus) have to interact with the MIP binding sites. A high flow rate could cause the bacteria to pass through the sensor surface too quickly, making it impossible for the MIP cavities to effectively capture them. Figure 13 shows the QCM sensor responses at stage C and D in an experiment with nebulized S. aureus bacteria. In the ~ 170th second, the air flow rate was adjusted to ~ 7 L/min. The sensors showed the large loading of the particles on them. But when the air flow rate was adjusted to ~ 9 L/min (in the ~ 225th second), some loaded particles became free. This observation verified the optimization result for optimum flow rate. Therefore, the flow rate was optimized at 7 L/min to strike a balance between collection efficiency and the mechanical stability of the captured bioaerosols.Fig. 13The QCM sensor response to flow rate adjustment. The QCM sensor response to flow rate adjustment. According to the pressure measurements (Table 9), increasing flow rate result in higher pressure drop, inducing mechanical stretching in microorganisms, potentially distorting their structure and preventing proper alignment with the imprinted cavities in the MIP layer. This mismatch may reduce the efficiency of bacterial capture. Consequently, reducing the flow rate can improve the performance of the MIP-based biosensor.Table 9Pressure drops in each stage.ΔP (mbar)Flow (L/min)ABCDETotal30.06970.07010.07270.35212.633.2040.06970.07020.07510.57183.824.2150.06970.07050.07810.85425.356.8060.06980.07080.08171.19937.239.0070.06980.07130.08611.60729.4511.25 Table 9Pressure drops in each stage.ΔP (mbar)Flow (L/min)ABCDETotal30.06970.07010.07270.35212.633.2040.06970.07020.07510.57183.824.2150.06970.07050.07810.85425.356.8060.06980.07080.08171.19937.239.0070.06980.07130.08611.60729.4511.25 Pressure drops in each stage. On the other hand, the bacterial cell wall shows viscoelastic behavior: Tensile strength (σ) is ≈30 MPa for Gram-Positive and ≈13 MPa for Gram-Negative (such as E. coli)26,27. The impact energy (Table 3) at all stages of the cascade impactor is below the rupture threshold for both Gram-positive and Gram-negative bacteria. Then the risk of mechanical damage to the target organisms is minimized. Finally, for evaluation of the specificity of MIP coating to capture the target bacteria, two types of MIP were coated on the sensor. One of them was with E. coli template and another one was based on the S. aureus template. These sensors were inserted at stage D of two identical cascade impactors. Figure 14 shows the sensors responses while nebulizing of S. aureus mixture. The sensor with S. aureus-MIP coating has very large frequency shift for its target microorganism (curve in blue). In contrast, the sensor with E. coli-MIP coating shows no variation for its non-target microorganism (curve in red).Fig. 14The QCM sensor response with various MIP coating. The QCM sensor response with various MIP coating. It is worth mentioning that exposure time is a critical parameter in biosensor performance. Insufficient exposure time may result in incomplete binding and reduced sensitivity. Long exposure time can increase the non-specific adsorption, without improving specific binding. Therefore, an optimized exposure duration (14.6 min), provides a practical balance between efficiency and specificity. Discussion This study examined glycan and MIP-based coatings with different functioning principles. As glycan-based coatings, glycoproteins like GA2 and mucin contain oligosaccharide chains that serve as identifying markers for proteins on the outer layer of cells that can be employed in microbe detection devices. These oligosaccharide chains are attached to the protein through either N- or O-glycosidic bond. But glycans are biomolecules that are extremely susceptible to environmental factors and can quickly lose their biological function over time, which presents challenges in the biosensor coatings. It is important to combine glycan-based elements with increased long-term stability into biosensors that allow the production of efficient, flexible, and low-cost analytical equipment28. Due to the low stability of glycan-based coating, the MIPs have garnered interest for bacterial detection in biosensor29. The MIP-based QCM sensor approach overcomes some limitations in other traditional methods through several mechanisms: Enhanced specificity and stability: MIPs can be engineered to mimic the specific binding sites of targets, providing targeted selectivity. Unlike antibodies, MIPs are robust and chemically stable. Cost-effectiveness: MIPs are cost-effective and suitable for mass production, making them ideal for broader clinical implementation. Label-free detection: the QCM sensor enables real-time, label-free detection without requiring secondary labels or reagents, eliminating complexity and additional costs. There are many relevant individual papers particularly on MIPs for pathogen detection30–32. A new study presents a novel cyclodextrin-integrated molecularly imprinted polymer (MIP)-based quartz crystal microbalance (QCM) sensor specifically designed for label-free detection of Galectin-3 (Gal3), a critical biomarker for heart failure progression16. The sensor architecture represents an integration of multiple molecular recognition strategies for a biomarker in the blood medium. Other recent studies focus on the integration of MIP-QCM for real-time detection of some substances in liquid medium such as milk and blood33–35. But none of these researches specifically address the integration of our three techniques together (MIP + QCM + cascade impactor), nor do they focus on their combined application for airborne pathogen monitoring. The SCI-QCM platform we developed, with its MIP coating, effectively monitors airborne bacteria in a reusable and size-selective way. Cascade impactor-type samplers actually size airborne particles based on their aerodynamic properties, which can be related to their deposition in the human lung. The physical size, shape, or density of a particle are part of its aerodynamic characteristics. This specification of the cascade impactor sampler can help to improve specificity of the device to a target microorganism. In the present study, dopamine was used as the functional monomer to change the QCM biosensor surface by forming a polydopamine layer. Early on the polymerization process, S. aureus cells were integrated into the PDA matrix, most likely through a combination of hydrogen bonding and ionic interactions between the dopamine molecules and elements of the bacterial surface. After the polymerization process, the template bacteria were eliminated, leaving behind imprinted cavities that were chemically and morphologically complimentary to S. aureus. Acetic acid was applied as a washing solution. Hydrogen interactions between the bacteria and polydopamine break when they are exposed to an acid aqueous solutions decrease non-covalent bonding. This coating provides an ideal platform for biosensing applications because of its exceptional mechanical durability, hydrophilicity, and biocompatibility. The hydrophilic PDA-MIP film was attached onto the QCM sensor, which then demonstrated significant affinity for the target bacterium limit. Also, the sensor exhibited greater selectivity for S. aureus compared to other non-template bacteria, including E. coli. The QCM biosensor recognized the target bacteria through electrokinetic attraction toward the MIP-functionalized sensor surface. When a suspension containing S. aureus was introduced, a significant decrease in the QCM resonant frequency was observed, indicating successful bacterial capture. These findings demonstrate the potential of PDA-based MIP coatings for extremely specific and reliable bacterial detection. Under alkaline conditions, dopamine polymerizes easily to form a uniform and coated film without the need for complicated fabrication processes or additional supporting materials36. Due to containing an amino group, dopamine can be connected to the bacteria via hydrogen bonding. Non-covalent actions such as hydrogen bonding are often used as it permits easy removal of the template from the polymer matrix and quick binding of templates to MIPs. MIP-coated sensors have been successfully produced using this technique (Yadav 2022). While the primary focus of this study was on the integrated system's performance, the specificity of the Polydopamine (PDA) MIP was validated using E. coli as a structural and biological competitor. Since E. coli (rod-shaped, Gram-negative) showed a negligible frequency shift compared to S. aureus (spherical, Gram-positive), it confirms that the binding is dictated by the stereospecific cavities created during the imprinting process. Actually, our selectivity tests using E. coli already provide a rigorous baseline. Furthermore, literature on PDA-based MIPs consistently reports an Imprinting Factor (IF) > 3.0, supporting the inherent selectivity of our protocol. It has been discussed in detail the impact of the surface characteristics (such as hardness, hydrophobicity, porosity, etc.) on the stability of bacterial adhesion37. Accordingly, the main characteristics for enhancing bacterial adhesion to surfaces are as follows: positive surface charge, suitable roughness (~ Ra = 200 nm), and appropriate stiffness (e.g., Young's modulus ~ 5 MPa). In the current study, it has also been shown, according to Table 10 and Figs. 7 and 11, that the MIP-based coating can meet these conditions.Table 10Sensor surface properties.Surface typeE, Young’s modulus (MPa)Ra, measured surface roughness (nm)QCM with MIP coating5158 ± 9Raw QCM50244 ± 10 Table 10Sensor surface properties.Surface typeE, Young’s modulus (MPa)Ra, measured surface roughness (nm)QCM with MIP coating5158 ± 9Raw QCM50244 ± 10 Sensor surface properties. Also, our method differs fundamentally from conventional Surface Plasmon Resonance (SPR)- and laser-induced fluorescence (LIF)-based bioaerosol detection in several key aspects. First, compared to SPR technique, QCM sensor measures mass changes directly via frequency shifts, whereas SPR detects refractive index changes near a metal surface. SPR typically requires a liquid phase and well-controlled optical alignment, making it less suitable for direct airborne, size-resolved sampling. In contrast, your system integrates aerodynamic size classification (via cascade impaction) with in situ mass sensing, which is advantageous for real airborne monitoring without liquid transfer steps. Second, compared to LIF methods, which detect intrinsic fluorescence signatures of biological particles (e.g., NADH, riboflavin), your method is selective at the molecular recognition level due to the molecularly imprinted polymer (MIP). LIF is rapid and real-time but generally non-specific and prone to false positives from non-biological fluorescent particles. Your MIP-QCM approach provides higher biochemical selectivity, particularly for target organisms such as Staphylococcus aureus. Third, LIF systems are excellent for high-throughput aerosol screening but do not provide direct quantitative mass accumulation on a functionalized surface. Your method combines aerodynamic fractionation with selective capture and direct gravimetric sensing, which enhances analytical specificity and enables potential sensor reusability. In summary, SPR offers high sensitivity in controlled liquid systems, LIF offers rapid but broad bioaerosol detection, while your integrated SCI-MIP-QCM platform provides size-resolved, selective airborne pathogen monitoring with stronger analytical specificity in real environmental conditions. Finally, a qualitative comparison with existing online bioaerosol detection technologies is summarized in Table 11. While optical methods such as LIF and SPR offer high sensitivities, they often require complex optical alignments or expensive reagents. Our system provides a balance between sensitivity and field-deploy ability, offering a cost-effective, label free solution for continuous monitoring in public spaces.Table 11Comparison with some technology.TechnologySensitivity (LOD)Real-timeCostMatrix robustnessThis work (MIP-QCM-SCI)ModerateYesLowModerateLIFHighYesHighHighSPRVery highYesHighHigh Table 11Comparison with some technology.TechnologySensitivity (LOD)Real-timeCostMatrix robustnessThis work (MIP-QCM-SCI)ModerateYesLowModerateLIFHighYesHighHighSPRVery highYesHighHigh Comparison with some technology. Sensitivity our device is 7.3 CFU/mL/min for bare sensor and about twice better for coated sensor (based on Fig. 9) and it is comparable with result of other researches38. To evaluate the sensor’s robustness in complex matrices (including PM2.5), we emphasize that the present system employs some selection mechanisms: (1) The cascade impactor acts as a high-precision physical pre-filter. It aerodynamically separates particles into five stages. The SCI serves as an "in-line sample preparation" unit. The SCI's aerodynamic stage C and D ensure that only particles within the 0.5–1.5 µm range reach the QCM surface. S. aureus has typically 0.5–1 µm diameter. The physical separation significantly mitigates the matrix effect, as larger fungal spores or smaller inorganic fumes are diverted to other stages A and B or E before they can interfere with the sensor. Figure 15 clearly shows the screening capability of this sampler, with surface D attracting the most S. aureus bacteria. This drastically reduces many "false particle load" on the QCM surface. (2) Unlike a bare QCM, our sensor is coated with a nano-thin MIP layer. While non-target particles may physically rest on the surface, they lack the specific shape and functional group complementarity required to "fit" into the imprinted cavities. (3) Mechanical entrapment of rigid inorganic particles often results in unstable or reversible frequency shifts, whereas the bio-recognition of S. aureus within the MIP cavities produces a characteristic, stable binding kinetic. By diverting the air flow from optimum flow, the other mechanically effect captured particles are removed and the target particle remain into the cavities.Fig. 15SCI aerodynamically separates particles into the stages (flow = 7 L/min). S. aureus predominantly is captured at stage D. SCI aerodynamically separates particles into the stages (flow = 7 L/min). S. aureus predominantly is captured at stage D. In addition, for very similar particles to S. aureus, some other tests were performed with mixture of S. aureus and Micrococcus. The amount of the captured bacteria on the stage D was 2.72 ± 0.76, 1.89 ± 1.99 CFU/mL respectively. It seems that the system is more sensitive to S. aureus to Micrococcus, but the difference is not statistically significant. It is worth mentioning that these bacteria have approximately same shape and diameter, then they cannot be differentiated by MIP + SCI combination. Conclusion S. aureus has become a serious global issue, especially with the increase in hospital-acquired infections and multidrug-resistant strains. Because of this, it’s really important to be able to identify and study this bacterium quickly, especially for biological and clinical purposes. Biosensors are gaining attention as affordable, easy-to-use, and fast alternatives for detection. Over time, many methods have been developed to find and measure harmful bacteria, like PCR, immunological tests, and traditional culturing and staining. Although these methods work well, they usually need trained experts, can be expensive, take a long time, and require a lot of work. Therefore, a small, real-time, sensitive, and cost-effective device is a need for pathogen monitoring. Many of the real-time, viable or bio-detection capable devices are high cost (Table 12). If present research can be commercialized, the device would be reliably capable to distinguish biologic vs non-biologic cheaply (about $10,000) and it can serve mid-tier market segments that cannot afford BioFlash or BioTrak but want better than just particle monitors. On the other hand, general particle counters are not able to capture, count and sort particles on surfaces when highly precise particle analysis is required.Table 12Some marketed devices’ costs.Product / systemDetection timePrice & cost structurePros / limitations relevant to present workTSI BioTrak (Dpstar Group)Near real‐time viable / inert discrimination; but does not do full species identification onboardVery expensive: unit cost listed around $67,500. Also, substantial operating / maintenance cost (filters, calibration, upkeep) though not all publicStrong technical capability; well established in pharmaceutical / clean room / regulatory environments. Our device competes on cost and simpler consumablesBioFlash-E (wmddetectorselector.army.mil)Reaction times tend to be quite fast. (2 min or less detection in some use cases; up to 21 biological threat agents)System cost ~ $35,000. Consumables (BioDiscs) cost maybe ~ $100 eachReal time with identification of specific pathogens. It is high cost, with consumables. Their market is more specialized (government / military / high security / first responders) rather than broad consumer or industrial Table 12Some marketed devices’ costs.Product / systemDetection timePrice & cost structurePros / limitations relevant to present workTSI BioTrak (Dpstar Group)Near real‐time viable / inert discrimination; but does not do full species identification onboardVery expensive: unit cost listed around $67,500. Also, substantial operating / maintenance cost (filters, calibration, upkeep) though not all publicStrong technical capability; well established in pharmaceutical / clean room / regulatory environments. Our device competes on cost and simpler consumablesBioFlash-E (wmddetectorselector.army.mil)Reaction times tend to be quite fast. (2 min or less detection in some use cases; up to 21 biological threat agents)System cost ~ $35,000. Consumables (BioDiscs) cost maybe ~ $100 eachReal time with identification of specific pathogens. It is high cost, with consumables. Their market is more specialized (government / military / high security / first responders) rather than broad consumer or industrial Some marketed devices’ costs. MIPs are becoming popular tools for recognizing specific molecules, acting somewhat like artificial antibodies to detect bacteria and viruses. In our study, we created a polymer sensor that can detect bacteria in environmental samples with good specificity and low cost. We made sure to imprint the exact surface structure of S. aureus onto the polymer, which allowed us to identify it specifically. Combining this MIP with a quartz crystal microbalance (QCM) lets us detect bacteria without labels in just a few minutes. Our sensing method uses both the size of the bacteria and their surface chemistry. Since our samples included both Gram-positive and Gram-negative bacteria, this method seems well suited for real-world environments with many different bacterial types. Its speed, accuracy, and ease of use make it promising for hospitals, biosafety labs, and industries sensitive to contamination. Overall, this system offers a strong and flexible approach for real-time detection of airborne bacteria. Theoretical models fit well with our experiments, helping us predict how particles stick or detach. Plus, the modular design makes it easy to regenerate and reuse, so it works well both in labs and out in the field some realistic limitations of our research might be improper selectivity in mixed aerosols, the flow might drift over time (because of cascade contamination), and the sensor will need clear regeneration/reuse cycles. Also, the high humidity condition (wet condition) can impact on the QCM performance. Then, in such environment, the integration of a dryer to the inlet of the device is necessary. A software-based differential baseline compensation algorithm (based on a humidity-temperature sensor that monitor the environment conditions) could enhance the detection precision. In the future, adding wireless readouts could allow remote monitoring and data sharing, while automation might improve efficiency and reduce mistakes. We’re also considering multiplexed receptor arrays that could detect multiple bacteria at once, covering a wider range of pathogens.