Pluronic F-127, potassium tetrachloroplatinate(II) (K₂PtCl₄), sodium tetrachloropalladate(II) (Na₂PdCl₄), L-ascorbic acid, acetone, potassium carbonate (K₂CO₃), sucrose, bovine serum albumin (BSA), 3,3′,5,5′-tetramethylbenzidine (TMB), TWEEN 20, hydrogen peroxide (H₂O₂), rabbit antisheep IgG antibody and pH-indicator strips were purchased from Sigma-Aldrich (St. Louis, MO). Hydrochloric acid (HCl) was obtained from Lab Alley (Austin, TX). Phosphate buffered saline (1× PBS, pH 7.4) was obtained from Caisson Laboratories (Smithfield, UT). Sheep anti-S-PMA polyclonal antibody and S-PMA-BSA were purchased from Chemitrace (Kendal, UK). Nitrocellulose (NC) membrane was purchased from Sartorius (Göttingen, Germany). Backing cards were purchased from DCN Dx (Carlsbad, CA). Glass fiber pads and cellulose absorbent pads were purchased from Millipore (Burlington, MA). The lateral flow assay cassettes were purchased from Drummond Scientific Company (Broomall, PA). Ultrapure water, obtained from a Millipore Milli-Q water purification system, was used for experiments.

Mesoporous Pd@Pt NPs were synthesized following a previously published method.⁴⁴ Briefly, 100 mg of Pluronic F-127 was first ultrasonically dissolved in a mixed solution containing K₂PtCl₄ (9 mL, 20 mM), Na₂PdCl₄ (1 mL, 20 mM) and HCl (220 μL, 6 M). Then, 10.0 mL of 100 mM L-ascorbic acid (AA) was added, and the mixture was subjected to continuous ultrasonic treatment at 35 °C for 4 h. The final product was centrifuged and washed five times with acetone and ultrapure water, respectively, before being dried at room temperature. After resuspending in 25 mL of ultrapure water, the synthesized Pd@Pt NPs, with a concentration of 2 mg/mL, were stored at 4 °C for further use.

Mesoporous Pd@Pt NP/antibody conjugates were prepared according to a previously published method with slight modifications.²⁸ The pH of the Pd@Pt NP suspension (2 mg/mL) was first adjusted to 8.2–8.5 by adding 0.02 M K₂CO₃. Subsequently, 5 μg of sheep anti-S-PMA polyclonal antibody was introduced into 1 mL of the adjusted NP suspension, followed by shaking incubation at room temperature for 60 min. Afterward, 10% (w/v) BSA was added, and the mixture was incubated with shaking for an additional 30 min. The conjugates were then collected by centrifugation at 9,000 rpm for 15 min and washed twice with 1× PBS containing 1% BSA. Finally, the mesoporous Pd@Pt NP/antibody conjugates were resuspended in 100 μL of PBS containing 2% BSA and 3% sucrose and stored at 4 °C for further use.

The LFIA consisted of five components: a sample pad, a conjugate pad, an NC membrane, an absorbent pad, and a backing card. The sample and conjugate pads, both made from glass fiber, were cut to specific sizes (300 × 20 mm for the sample pad and 300 × 5 mm for the conjugate pad), soaked in boric acid/borax buffer (pH 8.0), and then dried overnight at room temperature. The NC membrane was affixed to the backing card, and S-PMA-BSA (1.5 mg/mL) and rabbit antisheep IgG (1.0 mg/mL) were dispensed onto designated locations using an XYZ3050 Dispense System (BioDot Inc., Irvine, CA) to form the test and control lines, respectively. The membrane and backing card assembly were then dried at 37 °C for 3 h. Next, the sample pad, conjugate pad, and a cellulose absorbent pad were assembled onto the backing card, and the assembly was cut into 4 mm-wide individual strips using a CM4000 Guillotine Cutter (BioDot Inc., Irvine, CA). Subsequently, 2.0 μL of Pd@Pt NP/antibody conjugates were applied to the overlapping region of the conjugate and sample pads. Finally, each test strip was placed into a lateral flow assay cassette.

50 μL of the sample to be detected was mixed with 50 μL of PBST (mixed solution containing 0.1% TWEEN 20) running buffer and added to the conical sample port, where it migrated to the absorbent pad by capillary action. Once the black color appeared and stabilized on the test and control lines (approximately 10 min), the test strip was inserted into the cartridge slot of the 3D-printed device, and an image was captured using the smartphone. Next, 1.0 μL of TMB solution (containing 0.03% H₂O₂) was applied to the test line, and the appearance of a blue colorimetric signal was observed. A second image was then captured using the same setup. The readout results of the test strip were provided by the “Strip Scan” app.

For LFIA optimization, test strip images were captured using a HUAWEI Mate 30 (equipped with a 40-megapixel main rear camera) and analyzed with ImageJ (NIH, MD) to determine the peak area values of the test and control lines. To assess the detection performance of the proposed system (for S-PMA detection in PBST and urine samples), images of the test strips were captured and analyzed using our 3D-printed smartphone reader.

TEM measurements were conducted using a FEI Tecnai G2 20 Twin, operating in bright field mode with a 200 kV LaB6 electron source, and images were captured using an integrated 4K Eagle camera. SEM imaging was performed using an Apreo VolumeScope SEM equipped with a high-resolution secondary electron detector, operating under high vacuum. For optimal resolution and detail, the SEM images were obtained at a 30 kV accelerating voltage, 0.8 nA beam current, 3.6 mm working distance, and a 35 μs scan speed.

The proposed S-PMA detection system, as illustrated in Scheme 1, integrated LFIA with a 3D-printed smartphone reader. The LFIA functions as the core component of the S-PMA detection and operates on the principle of competitive assays. In the absence of S-PMA in the test sample, the applied Pd@Pt NP/antibody conjugates migrate along the NC membrane to the test line. There, they bind to S-PMA-BSA, forming Pd@Pt NP/antibody-S-PMA-BSA complexes, which result in a visible black line. The unbound Pd@Pt NP/antibody conjugates continue migrating to the control line, where they bind to rabbit antisheep IgG, forming another black line. When S-PMA is present in the test sample, it competes with S-PMA-BSA on the test line for binding to the Pd@Pt NP/antibody conjugates. This competition reduces the formation of Pd@Pt NP/antibody-S-PMA-BSA complexes at the test line, resulting in a gray line instead of a black one. Upon addition of TMB and H₂O₂, the peroxidase-like catalytic activity of the Pd@Pt NPs produces a strong blue color, enhancing the colorimetric signal. The quantification of LFIA results was achieved using the proposed 3D-printed smartphone reader, where the 3D-printed device ensured vibration resistance, accurate positioning, and improved contrast for image capture, while the smartphone app enabled on-site data analysis.

To evaluate the morphology of the synthesized Pd@Pt NPs, we performed scanning electron microscopy (SEM) and transmission electron microscopy (TEM), as shown in Figure 1B. The Pd@Pt NPs exhibited a spherical, hyperbranched, mesoporous nanostructure with a uniform size of 45–50 nm, which was evident in both SEM and TEM images. This distinctive concave morphology resulted from the formation mechanism of mesoporous bimetallic Pd@Pt NPs, described as a surfactant-assisted self-assembly process, as illustrated in Figure 1A. In this process, the Pluronic F-127 surfactant first formed spherical micelles, with hydrophobic poly(propylene oxide) (PPO) positioned internally and hydrophilic poly(ethylene oxide) (PEO) oriented externally, serving as structural templates for the nanoparticles. With the introduction of ascorbic acid (AA) as a reducing agent, a Pd-dominated core initially formed, followed by the subsequent deposition of Pt, resulting in a core–shell structure. This sequential growth was driven by the faster reduction rate of the Pd precursor compared to that of the Pt precursor. The high lattice match between Pd and Pt led to the epitaxial growth of Pt, promoting the formation of the uniform concave morphology. Additionally, the presence of HCl during synthesis optimized the pH conditions for Pt growth, resulting in the highly branched Pt structures in the shell. Simultaneously, F127 micelles, in coordination with metal-aqua complexes, served as structure-directing agents.⁴⁵ From the TEM image (Figure 1B), the Pt dendrites (2–3 nm in size) that construct the outer shell, supported by the Pd inner core, were distinctly visible. This unique mesoporous, dendritic, flower-shaped core–shell structure provides a large surface area and numerous active sites for catalytic reactions, contributing to the high catalytic performance of the Pd@Pt NPs.

Following antibody conjugation, TEM (Figure 1C) image shows that the NPs preserved their spherical, mesoporous, hyperbranched morphology and consistent size, suggesting that the conjugation process did not compromise the structural integrity of the particles. Additionally, high-contrast TEM image revealed a faint halo around the particles, indicating successful antibody attachment to the Pd@Pt NP surface. The absence of structural deformation or aggregation in the TEM image further confirmed the stability of the Pd@Pt NPs after conjugation.

To achieve optimal LFIA performance, several key parameters were optimized (as shown in Figure 2), including the amount of conjugated antibody in the preparation of Pd@Pt NP/antibody conjugates, the volume of Pd@Pt NP/antibody conjugates, the concentration of S-PMA-BSA on the test line, and the pH of the phosphate-buffered saline containing 0.1% Tween-20 (PBST) running buffer. The amount of conjugated antibody and the volume of Pd@Pt NP/antibody conjugates influence the binding efficiency of the detection probe to the target, thereby affecting the performance of the LFIA. As shown in Figure 2A, the LFIA signal increased and reached its maximum at 5 μg under both conditions (with or without TMB and H₂O₂), then decreased at 6 μg, indicating that the optimal result was achieved by adding 5 μg of S-PMA antibody to 1 mL of Pd@Pt NPs (2mg/mL). As presented in Figure 2B, for varying volumes of Pd@Pt NP/antibody conjugates, the signal increased up to 2.0μL and remained stable (with TMB and H₂O₂ added), indicating saturation of the blue colorimetric signal. Without TMB and H₂O₂ added, the signal decreased at 2.5 μL. Therefore, 2 μL of Pd@Pt NP/antibody conjugates was determined to be optimal for both conditions. The S-PMA-BSA on the test line competes with the target for binding sites on the Pd@Pt NP/antibody conjugates. The concentration of S-PMA-BSA affects the signal modulation of LFIA detection and, thus, determines the detection range. As shown in Figure 2C, the signal reached its maximum at 1.50 mg/mL under both conditions (with or without TMB and H₂O₂). Therefore, 1.5 mg/mL of S-PMA-BSA was selected for optimal results. For LFIA detection, the PBST running buffer is critical for minimizing nonspecific interactions. The pH of PBST can also affect the binding efficiency of the detection probe to the target, as well as the peroxidase-like catalytic activity of Pd@Pt NPs, thereby influencing the overall performance of the proposed LFIA. A pH close to physiological conditions (6.5–7.9) is generally ideal for optimal binding. As shown in Figure 2D, signal variation was minimal for both conditions (with or without TMB and H₂O₂), and the optimal result was obtained at pH 7.1. In summary, 5 μg of conjugated antibody, 2 μL of Pd@Pt NP/antibody conjugates, 1.50 mg/mL of S-PMA-BSA, and PBST running buffer at pH 7.1 were selected for subsequent experiments.

To enhance image contrast and enable quantification of the test strip results, we designed a 3D-printed smartphone-based lens device, as shown in Figure 3. The device consisted of three main components: a smartphone clamp, a major block housing strip cartridge slots, and a lens holder with a convex lens. These components were designed in Fusion 360 (CAD software) and fabricated using a commercial 3D printer (Einstar3D, Co., Shenzhen, China).

The choice of white polylactic acid (PLA) filament and the hollow design of the major block facilitated optimal ambient light transmission for image capture. To accommodate smartphones with varying thicknesses and camera positions, an adjustable phone clamp was employed. Instead of a holder with specific dimensions for specific smartphone model, this phone clamp design can spread the applicability of the 3D-printed device. The designed phone clamp was expected to suit majority of the manufactured smartphones in the market. An easily removable and adjustable lens holder with a convex lens was placed in front of the strip cartridge slot within the major block. This additional lens expands the field of view (FOV) when capturing images at close distances using a smartphone, which enhances image contrast by improving resolution and reducing scattered ambient light.

Figure 3B shows the images taken with (right) and without (left) the lens device, demonstrating the improved contrast when using the 3D-printed apparatus. Notably, optical lenses with different focal lengths can be selected and installed on the lens holder, and its y and z positions can be adjusted to fit various smartphone camera setups. Additionally, the 3D-printed smartphone-based lens device enables image capture with vibration resistance, making it suitable for on-site applications that require stable and accurate measurements under diverse ambient conditions. The entire 3D-printed device is compact, with dimensions under 6 cm in length, 4 cm in width, and 4 cm in height, and costs less than $2.

Images of the test strips were captured and analyzed using our 3D-printed smartphone reader, as demonstrated in Figure 3C and 3D. A previously developed Android strip reader app, “Strip Scan”,⁴⁰ was employed to generate quantitative results from the captured images. The typical interface of the app is shown in Figure 3D, where the “FIRST” and “SECOND” peak area values were extracted to quantify the test strip results.

To evaluate the performance of the proposed detection system, varying concentrations of S-PMA in PBST were tested and compared to results from Au NP-based standard LFIAs using previously optimized parameters. Figures 4A, 4B, and 4C depict the LFIA detection of S-PMA at concentrations of 1000, 100, 10, 1, 0.1, 0.01, and 0 ng/mL (from left to right), utilizing Pd@Pt NPs, Pd@Pt NPs with the addition of TMB and H₂O₂, and Au NPs, respectively. Quantitative results were obtained by capturing images using the 3D-printed device, cropping the test zones, and analyzing them with a previously developed smartphone app that employed a novel algorithm to extract the peak area values of the test lines.

Due to the principle of competitive LFIA, the signal, represented by the peak area value of the test line, is expected to be inversely proportional to the concentration of S-PMA. Figure 4D shows the relationship between the peak area values and the log-transformed concentrations of S-PMA (0, 0.01, 0.1, 1, 10, 100, and 1000 ng/mL) in PBST for all three LFIAs, which confirmed the inverse correlation. The linear detection range for the standard LFIA using Au NPs as signal labels was 1–1000 ng/mL (y = –1595.6x + 4836.6, R² = 0.98). Similarly, the LFIA using Pd@Pt NPs without TMB and H₂O₂ showed a comparable linear range of 1–1000 ng/mL (y = −1709.6x + 7927.5, R² = 0.977). Notably, the addition of TMB and H₂O₂ extended the linear range to 0.1–1000 ng/mL (y = −1301.1x + 8474.6, R² = 0.9818), with the lower boundary decreased by 1 order of magnitude.

We defined the limit of detection (LOD) as the concentration of S-PMA at which the signal intensity equals the mean signal intensity at 0 ng/mL minus three times the standard deviation. The LOD of the Pd@Pt NP-mediated LFIA, in the absence of TMB and H₂O₂, was calculated to be 6.96 ng/mL, which is lower than the calculated LOD of the Au NP-based LFIA (9.72 ng/mL). With the addition of TMB and H₂O₂, the LOD decreased to 0.50 ng/mL, showing a one-order-of-magnitude reduction. This significant enhancement in sensitivity was attributed to the peroxidase-like catalytic activity of mesoporous Pd@Pt NPs, which acted as signal amplifiers in LFIA detection. Because of the peroxidase-like property of Pd@Pt, it can produce an amplified colorimetric signal in the presence of TMB + H₂O₂, which is the sum of the inherent colorimetric signal of the nanozyme and the catalytic product, consequently, enhancing the sensitivity of LFIA.²⁴

To demonstrate the proof-of-concept and potential application for the practical screening of human exposure to wood smoke, human urine samples were spiked with varying known concentrations of S-PMA and subsequently analyzed using the proposed detection system. Figure 5 shows the relationship between the peak area values and the logarithm of S-PMA concentrations (0.5, 1, 2, 4, 20, 100, and 500 ng/mL) in urine samples. The linear range for the LFIA, in the absence of TMB and H₂O₂, was 2–500 ng/mL (y = −3241.7x + 9140.2, R² = 0.9936). When TMB and H₂O₂ were added, the linear range expanded to 0.5–500 ng/mL (y = −3331.8x + 9715.8, R² =0.9948). Notably, the calculated LOD for the Pd@Pt NP-mediated LFIA with TMB and H₂O₂ was 2.50 ng/mL, which is considerably lower than the LOD of the LFIA without TMB and H₂O₂ (7.05 ng/mL).

The biological exposure index (BEI) for urinary S-PMA levels of 25 μg/g creatinine is suggested by the American Conference of Governmental Industrial Hygienists (ACGIH) as biomonitoring equivalent to an exposure of 0.5 ppm benzene in the air.⁴⁶ Since typical creatinine concentrations in human urine range from 80 to 200 μg/dL,⁴⁷ the LOD of our proposed detection system for S-PMA is substantially lower than the ACGIH recommended standard. This highlights the practical significance of our method for assessing human exposure to wood smoke.

Additionally, we calculated the coefficient of variation (CV) for the peak area of ten urine samples spiked with 20 ng/mL and another ten spiked with 100 ng/mL S-PMA, as shown in Table 1. The CV values were relatively low under both conditions (with and without TMB + H₂O₂) and for both S-PMA concentrations (20 ng/mL and 100 ng/mL), demonstrating the excellent stability and reproducibility of our method.