Pollen is the male reproductive cell of plants and varies greatly in size, shape, and structure due to species differences, typically measured in micrometers. It can be classified into two types based on dispersal methods: entomophilous (insect-pollinated) and anemophilous (wind-pollinated) flowers. Entomophilous flowers rely on insects for pollen distribution and often have larger pollen grains produced in smaller quantities, primarily from colorful plants like peach blossoms and roses. However, the main contributors to pollen allergies are anemophilous flowers, which disperse their pollen through the wind [8].

In urban areas like Beijing, allergenic pollen predominantly comes from tree species such as cypress, pine, poplar, willow, and elm, as well as grasses like ragweed and mugwort. Tree and grass pollen are typically light and small, allowing them to travel long distances, when carried by strong winds. Therefore, as long as you are outdoors, especially in the spring when the flowers are in full bloom, it is almost inevitable to be exposed to pollen outdoors.

The diameter of pollen grains generally ranges from 30 to 50 micrometers [9]. When they disperse in the air, they can easily be inhaled into the human respiratory tract. Once pollen grains enter the respiratory tract and penetrate the mucous membranes, the body’s macrophages recognize the pollen allergen proteins as “invaders” and initiate an immune response against them. At this point, plasma cells in the body produce a large amount of specific antibodies. Antibodies are immunoglobulins with immune functions secreted by human cells in response to “invasion” stimuli, and they can specifically bind to antigens. These antibodies then attach to the surface of mast cells, sensitizing them to the allergen for future secondary exposure. Upon this re-exposure, the allergens cross-link the IgE antibodies on the mast cell surface, causing the cell to release histamine & other mediators, which triggers the characteristic allergy reaction. The whole process can be shown in Figure 1.

Figure 1. Schematic diagram of the human allergy process.

According to the structures, immunoglobulins are divided into IgG, IgA, IgM, IgD, and IgE [10], as shown in Figure 2. There are two types of antibodies involved in pollen allergy: IgE and IgG antibodies. IgE antibodies are indicators of the degree of allergy in the human body.

Elevated levels of IgE are signs of allergic diseases such as asthma, allergic rhinitis, and atopic dermatitis. In the pathogenesis of allergic diseases, IgE antibodies bind to high-affinity receptors on the surfaces of mast cells and basophils, triggering these cells to release inflammatory mediators and leading to allergic symptoms. IgG antibodies are the most abundant antibodies in the human body. The Fab fragment (antigen-binding fragment) of IgG can specifically recognize and bind to epitopes on antigens. This specific binding capacity allows IgG antibodies to distinguish between different antigens and only bind to specific targets. This specificity is the fundamental reason why the chemically modified graphene sensor in this research can differentiate pollen proteins from other impurities and distinguish between different types of pollen proteins.

Figure 2. Illustration of five antibodies in human body.

Graphene’s high sensitivity in detection applications stems from its unique features of 2D structure, large surface area, and high carrier mobility, allowing it to detect subtle changes from biomolecular binding, doping or gas adsorption, enabling early disease diagnosis, environmental monitoring, and point-of-care devices with limits down to femtomolar (10⁻¹⁵) levels. Its atomic thinness exposes all atoms, amplifying interactions, and its biocompatibility and strong π-π interactions further enhance its sensing capabilities.

Figure 3. Schematic diagram of graphene doping types.

Graphene’s high sensitivity in detection originates from its 2D structure, large surface area, and high carrier mobility, with its π-conjugated zero-bandgap structure making it sensitive to surface charge transfer that induces n-type or p-type doping (n-type: electron gain, reduced resistance; p-type: electron loss, increased resistance, more holes), as is shown in Figure 3. This research’s p-type doping mechanism induced by allergen protein binding hinges on frontier molecular orbital (FMO) theory [11] and structural changes in the antibody-antigen complex, which strengthens the link between the biological binding event and the electronic signal.

The core principle for the protein-antibody complex acting as a “hole acceptor” lies in the relative energy levels of graphene and the combination. Intrinsic graphene has a Fermi level aligned with its Dirac point, where the valence band and conduction band meet [12]. When IgG antibodies bind to pollen allergen proteins (e.g., Bet v1, Amb A1), three key changes occur in the combination:

1) Energy Level Realignment: The formation of the immunocomplex results in a significant shift of its lowest unoccupied molecular orbital (LUMO). When this LUMO energy level drops below the graphene, The binding event shifts the LUMO to an energetically accessible level beneath the graphene Fermi level, establishing a gradient that drives electron transfer from the substrate to the complex.

2) Enhanced dipole moment: Structural rearrangements within the antibody’s Fab fragment upon binding reorient polar residues, increasing the complex’s molecular dipole moment. This enhanced dipole acts as a local electrostatic gate, further modulating the carrier density of the graphene [13].

3) Thermodynamic Driving Force: The stability of the resulting antibody–antigen bonds lowers the system’s Gibbs free energy, providing the driving force for the charge transfer process [14].

According to Frontier Molecular Orbital (FMO) theory, this electron withdrawal from graphene increases the hole carrier density, effectively inducing p-type doping. This process generates additional scattering centers at the interface, leading to a measurable increase in electrical resistance. This causal chain ensures that the resistivity change is highly specific to the intended biological binding event rather than non-specific interactions.

2.1.1. PBASE Modification

Using 1-pyrrolidine butyric acid N-hydroxysuccinimide ester (PBASE) to fix protein antibodies in pollen allergens to the graphene surface is one of the common methods of graphene chemical modification [15]. PBASE is a chemical modifier with a molecular formula/molecular weight of C₂₄H₁₉NO₄ = 385.42 g/mol [16], as is shown in Figure 4. PBASE has the properties of a surfactant, and its molecule contains an N-hydroxysuccinimide ester functional group, which reacts chemically with specific groups in antibodies to form stable covalent bonds, making it easier for allergen antibodies to be fixed on the graphene surface [17].

To construct a detection interface, we need to modify the linker PBASE first onto the surface of the graphene so that the antibodies can be fixed.

In this experiment, a 2mM methanol solution of PBASE was first prepared. The 0.0077084 grams of solid PBASE was weighed and dissolved in 10ml of methanol. Subsequently, the entire graphene substrate was immersed in the PBASE methanol solution. The graphene was soaked and modified in the PBASE solution for 1 hour at room temperature to functionalize the graphene surface.

After the modification, the graphene substrate was removed and washed three times in clean methanol to remove any unbound PBASE. Then the graphene was immersed in deionized water for 5 seconds to remove residual methanol. After that, the graphene was dried using a nitrogen gun to ensure that there was no solvent residue on the graphene surface. Finally, the functionalized graphene was stored at 3˚C for subsequent use.

In the experimental steps, special attention should be paid to the removal of unbound PBASE, which is a lesson learning from many experiments. In this process, the methanol and deionized water should be used to remove unconnected PBASE linker. In some experiments, due to insufficient cleaning, multiple layers of PBASE linkers were formed on the graphene surface, adding unnecessary height.

2.1.2. IgG Antibody Modification

In order to modify IgG antibodies, a 250 μg/mL IgG antibody/Phosphate-buffered saline (PBS) solution is required. For ease of calculation, the modified IgG solutions are prepared in units of mass concentration, which also avoids the complexity of converting to molar concentration, laying the foundation for subsequent research and calculation.

During the experimental process, to accurately modify two graphene samples, a total of 80 μL of 250 μg/mL birch pollen antibody/PBS solutions were required. This experiment utilized a pipette to achieve the goal of precise pipetting.

Experimental steps are as follows:

The antibody/PBS solution was first prepared by adding 10 µL birch pollen specific IgG antibody (3B4 anti-Bet v1) to 70 µL PBS and mixing them evenly. Then, 10 μL of antibody solution was added to the graphene surface. After that, the sample was placed in a 4˚C environment for modification for 4 hours. After modification, the sample was gently rinsed with PBS and deionized water and then was dried using nitrogen (N₂). Finally, the modified antibody was stored at 2˚C - 8˚C for subsequent use. The experiment showed that there is no significant decrease in antibody activity within one month.

2.1.3. Pollen Allergen Modification

Given that there is currently no research on the extraction solution of pollen allergen proteins detected by graphene, there is no existing experience or literature to refer to for the concentration ratios of the modified pollen allergen extract. In order to facilitate the analysis of different concentration specificities of allergen proteins, this experiment attempted in three different ratios: 1:5, 1:10, and 1:15. The ragweed pollen allergen extract with a concentration of 1.85 mg/5ml and the birch pollen allergen extract with a concentration of 2.75 mg/5ml were diluted accordingly to meet different needs.

Below are the concentrations calculated after three dilution ratios:

Ragweed pollen allergen:

•1:5 dilution: 0.3083 mg/ml;

•1:10 dilution: 0.1682 mg/ml;

•1:15 dilution: 0.115625 mg/ml.

Birch pollen allergen:

•1:5 dilution: 0.4583 mg/ml;

•1:10 dilution: 0.25 mg/ml;

•1:15 dilution: 0.171875 mg/ml.

Steps for modification:

First, prepare different concentrations of ragweed pollen allergen protein (AmbA1) and birch pollen allergen protein (Bet V1) in a pH 7.4 PBS buffer solution. Then, add droplets of these allergen protein solutions onto the modified graphene-PBASE-antibody structure. Next, place the samples at 4˚C for 30 minutes to allow for the allergen protein to antibody binding reaction. After the modification is complete, gently rinse the modified graphene samples with PBS and deionized water to remove any unbound allergen proteins. Finally, dry the samples with nitrogen gas (N₂) to complete the modification experiment.

2.1.4. Rationale for Key Experimental Parameters

Given the lack of literature precedents for graphene-based pollen allergen detection, the pollen allergen extract dilution ratios were determined through systematic preliminary optimization experiments (control variable design, n = 3 replicates per condition) to ensure reproducibility and performance.

The screening criteria included: 1) stability of graphene functionalization (via AFM characterization); 2) signal-to-noise ratio (SNR) of resistance changes; 3) absence of non-specific aggregation (via optical microscopy/AFM); and 4) consistency of antibody-antigen binding.

Details are as follows:

2 mM PBASE methanol solution

We tested 1 mM, 2 mM, and 3 mM PBASE solutions. The 1 mM PBASE resulted in insufficient functionalization, with AFM showing less than 50% surface coverage of PBASE protrusions, leading to weak antibody anchoring and inconsistent resistance signals. The 3 mM PBASE caused multilayer aggregation (10 - 15 nm protrusions), which disrupted graphene’s conductivity. In contrast, 2 mM achieved uniform monolayer coverage (0.5 - 3 nm protrusions) and stable covalent bonding with IgG antibodies, ensuring reliable antibody immobilization without interfering with charge transport.

250 μg/mL IgG antibody/PBS solution

We tested 100 μg/mL, 250 μg/mL, and 500 μg/mL concentrations. The 100 μg/mL resulted in sparse antibody immobilization (no detectable aggregates via optical microscopy) and negligible resistance changes (ΔR < 200 Ω, SNR < 3:1). The 500 μg/mL caused severe dendritic aggregation (1 - 2 μm), blocking charge transfer. In contrast, the 250 μg/mL provided a balanced dispersion of small particles and effective binding, with antibody activity retained for one month at 2˚C - 8˚C (verified by repeated binding assays).

The 1:5, 1:10, 1:15 allergen dilution ratios

We tested allergen dilution ratios of 1:20, 1:15, 1:10, 1:5, and 1:2. The 1:20 dilution (ragweed: 0.0925 mg/mL; birch: 0.0825 mg/mL) produced weak signals (ΔR < 200 Ω, SNR < 3:1) due to insufficient binding. The 1:2 dilution caused surface saturation (AFM revealed aggregated allergen-antibody complexes), leading to non-linear resistance responses. Ratios of 1:5 to 1:15 yielded ΔR values of 250 to 1900 Ω (SNR > 10:1), allowing for clear differentiation between concentrations while avoiding saturation.

2.2.1. Characterization of Prepared Graphene

Figure 5. Graphene observed under an optical microscope.

To make sure the prepared graphene is of good quality, a graphene characterization process is needed. Place the graphene on the glass slide under an optical microscope for observation. In the observation field, the blue block is graphene. The more purple the color is, the closer it is to the color of the substrate silicon wafer, making it more difficult to identify. This indicates that the graphene is thinner. By observing in this way, you can find the graphene with better quality, as is shown in Figure 5.

Figure 6. Graphene Raman spectroscopy.

Through Raman spectroscopy observation in Figure 6, it was found that the intensity of the G’ peak of the graphene is greater than the G peak, indicating a thinner layer, typically a single-layer or few-layer graphene [18]. At the same time, there are also certain defects, as reflected in the D peak.

2.2.2. PBASE Characterization after Modification

The following Figure 7 is the surface morphology scanned by atomic force microscopy after PBASE modification.

Figure 7. AFM height image of PBASE-modified graphene (left) and the corresponding line-scan height profile (right).

From the above AFM height image and height profile of the graphene surface, it can be seen that a large number of nanoscale protrusions (white bright spots, height 0.5 - 3 nm) appear on the graphene surface after PBASE modification. These protrusions represent covalently bonded PBASE molecules and their aggregates. The larger bright spots in the AFM height image (~5 - 10 nm) may be PBASE multilayer aggregates. Due to the fully covered monolayer graphene, there are no direct atomic-level steps, and the surface undulations (Figure 7 right image Root Mean Square roughness about 1 - 2 nm) should mainly result from the effects of PBASE molecular layer modification. By taking a segment along a distance on the image, the surface undulations in Figure 7 right image can be observed, confirming the result of PBASE physical adsorption modification. Based on the atomic force microscopy scan above, it can be concluded that PBASE has successfully modified the graphene surface.

2.2.3. Antibody Characterization after Modification

Figure 8. Modified antibodies observed under an optical microscope (high density left, low density right).

The two images in Figure 8 were taken under an optical microscope. In the left image, the graphene surface is observed to be covered with a large number of aggregates that are significantly thicker, densely interwoven, and have a dendritic and mountain-like structure, with lateral dimensions reaching 1 - 2 µm. This is the macroscopic aggregation of excess IgG antibodies during the drying process, suggesting that the initial IgG modification concentration was too high.

Figure 9 is a control image with low-concentration antibody modification, where only sporadic small particles can be seen on the surface without any large-area accumulation, confirming that reducing the amount of antibody can effectively inhibit the formation of aggregates.

Figure 9. AFM Characterization of antibody doping.

Due to the high concentration of antibodies in Figure 8 left, it is necessary to appropriately adjust the amount of antibody applied in the subsequent experiments. Figure 9 shows the image obtained by atomic force microscopy (AFM) scanning after reducing the amount of IgG antibodies applied. In the image, some coral-like antibody aggregates can still be seen, but in the next layer, uniformly distributed monolayer small antibodies are present. Upon close observation, each antibody appears Y-shaped [10].

The observations through optical microscopy and atomic force microscopy confirm that the antibodies have been successfully modified on the graphene surface, forming a graphene-based biosensor detection and monitoring interface.

2.2.4. Pollen Allergen Proteins Characterization after Modification

In this study, the binding of pollen allergen to antibodies was detected by monitoring the resistance and resistivity. However, in order to confirm that the impurities are pollen proteins, other methods are needed for supplementary analysis.

Figure 10. Pollen allergen protein modified AFM characterization.

The molecular weight of the birch pollen allergen Bet v1 protein is approximately 17 kDa [19], which is only one-tenth of that of a IgG (≈150 kDa) [20], with a crystallographic size of about 3 nm. The nearly spherical particles in the AFM images in Figure 10 have a height of 2 - 4 nm and a diameter of 8 - 12 nm, corresponding to an adsorbed layer mainly consisting of monomers and a few dimers, with no obvious aggregation observed. This result confirms that the birch pollen allergen protein Bet v1 has been successfully and uniformly modified on the detection interface of the graphene biosensor. Next, the doping effects of graphene caused by the above series of chemical modifications will be detected through resistance measurement.

2.3.1. Resistance Experiment Design

After a series of chemical modifications to graphene, we are now at the stage of experimental verification. This study designed two groups of three experimental processes to conduct 15 comparative experiments, with the goal of measuring the resistance changes of different combinations, calculating the resistivity changes before and after doping, and verifying the specific recognition and concentration recognition functions of the graphene sensor.

1) Experimental Group 1: On two graphene devices modified with antibodies, the corresponding specific pollen allergen and non-specific pollen allergen proteins were doped respectively, and the resistance changes of the two sensors before and after chemical modifications were measured to verify the specific detecting function of the graphene device. Three sets of repeated experiments were performed.

2) Experimental Group 2: On the graphene device modified with antibodies, the corresponding specific pollen allergen proteins of different concentrations were doped, and the resistance changes of the two devices before and after chemical modification were measured to verify the function of the sensor in detecting pollen allergen proteins of different concentrations. Three sets of repeated experiments were performed.

2.3.2. Resistance Measurement

1) Experimental Group 1:

a) Graphene sensors modified 250 μg/mL Ragweed pollen antibody 5F6 anti-Amb A1) doped with 1:5 diluted Ragweed pollen allergen Amb A1 solution was used for specific pollen protein detection, and three sets of experiments were repeated, as is shown in Table 1.

Table 1. Resistance comparison for specific pollen allergen detection before and after doping with allergen Amb A1 (1:5 dilution).

b) Graphene sensors (modified with 250 μg/mL Ragweed pollen antibody 5F6 Anti Amb A1) doped with 1:5 diluted birch pollen allergen Bet v1 solution, and non-specific pollen protein recognition detection was performed, and the experiment was repeated three times, as is shown in Table 2.

Table 2. Resistance comparison for non-specific pollen allergen detection before and after doping with allergen Amb A1 (1:5 dilution).

2) Experimental Group 2:

a) Three graphene sensors (modified with 250 μg/mL Ragweed pollen 5F6 Anti Amb A1) were doped with ragweed pollen Amb A1 solution diluted at 1:5, 1:10, and 1:15, respectively. The pollen allergen proteins of different concentrations were tested, as is shown in Table 3.

Table 3. Resistance comparison for specific pollen allergen detection before and after doping with different allergen Amb A1 dilution (1:5, 1:10, 1:15).

b) Three graphene biosensors (modified with 250 μg/mL Ragweed pollen 5F6 Anti Amb A1) were doped with Ragweed pollen Amb A1 solution diluted 1:5, 1:10, and 1:15, respectively. The pollen allergen proteins of different concentrations were tested, as is shown in Table 4.

Table 4. Resistance comparison for specific pollen allergen detection before and after doping with different allergen Amb A1 dilution (1:5, 1:10, 1:15).

c) Three graphene biosensors modified with 250 μg/mL Ragweed pollen 5F6 Anti Amb A1) were doped with Ragweed pollen Amb A1 solution diluted 1:5, 1:10, and 1:15, respectively. The pollen allergen proteins of different concentrations were tested, as is shown in Table 5.

Table 5. Resistance comparison for specific pollen allergen detection before and after doping with different allergen Amb A1 dilution (1:5, 1:10, 1:15).

2.3.3. Resistivity Calculation

Resistivity can help differentiate the conductivity properties of different materials, and even if two devices have the same resistance, their resistivity may be different.

Converting resistance to resistivity helps eliminate errors caused by different electrode lengths and cross-sectional areas between different graphene sensors.

Resistivity calculation formula:

where:

ρ is the resistivity (Ω⋅m);R is the resistance (Ω);A is the cross-sectional area of the conductor (m²);L is the length of the conductor (m).

Determinationofaverageheightofgraphenedoping

Figure 11. The modification process.

The average height of the functionalized graphene structure was quantified via atomic force microscopy (AFM) height profiling across all 15 sensors:

1) For each sensor, 10 random line scans (500 nm length, 0.5 nm resolution) were acquired using tapping-mode AFM.

2) Height was measured from the SiO₂ substrate to the top of the functionalized layer (graphene + PBASE + antibody + allergen).

3) Single-layer graphene contributes ~0.34 nm, PBASE adds 0.5 - 3 nm, IgG antibodies (Y-shaped) add ~15 nm, and pollen allergens (e.g., Bet v1) add 2 - 4 nm.

Considering all above heights together through the whole process of modification, as is shown in Figure 11, the average height across all line scans and sensors was around 20 nm, namely 2.0 × 10⁻⁸ m, accounting for minor variations in molecular layer thickness.

Table 6. Graphene sensors’ height and cross-sectional areas after doping.

The cross-sectional area includes the entire functionalized structure because charge transport in the sensor depends on the graphene’s interaction with the bound molecular layers (PBASE, antibody, allergen). Excluding these layers would underestimate the conductive volume, leading to artificially high resistivity values. Using the average height ensures consistency across sensors, as individual thickness variations (±2 nm) were negligible compared to the total height (~20 nm) and did not affect the statistical significance of results. Table 6 is the summary of average height cross-sectional areas after doping.

1) In Experimental Group 1, the resistivity data were obtained through the measurement and resistivity calculation experiments, as is shown in Table 7:

Table 7. Graphene sensors’ resistivity calculation.

No T The significance test is conducted to analyze the experimental data:

Null hypothesis (H₀): Resistivity change value after doping = Resistivity change value before doping.Alternative hypothesis (H_a): The resistivity change after doping is greater than the resistivity changes before doping.

The data has been numerically amplified (multiplied by 10⁶) to avoid calculation accuracy problems caused by too small values. The specific statistical description and T-test analysis result are shown in Table 8 and Table 9:

Table 8. Statistical analysis.

Table 9. T-test: Paired two-sample for means.

Test results: Degrees of freedom (df): 2; t-statistic (t Stat): 28.1347; Single tail p-value (P (T<=t) one-tailed): 0.00063.Significance analysis: p-value (0.00063) is much smaller than the significance level α = 0.05. Therefore, we reject the null hypothesis (H₀) and accept the alternative hypothesis (H_a).Conclusion: At a significance level of α = 0.05, the one-tailed t-test results indicate that the change in resistivity after doping is significantly greater than the change in resistivity before doping. This suggests that doping has a significant impact on the resistivity of the graphene sensor, validating the effectiveness of the experiment.

2) In Experimental Group 2, the resistivity data obtained from measurements and resistivity calculations are as follows. By visualizing the data, a trendline fitting is conducted to explore the relationship between pollen concentration and resistivity change. The resistance measurement is shown in Table 10.

Table 10. Resistivity for Experiment Group 2.

By using trend lines to fit the sample data points, three exponential function fitting lines were obtained. The R² value is close to 1, indicating a good fit. By combining the three lines, an exponential equation was derived as Figure 12:

y = 0.594e^−239.3x

Figure 12. The relationship between pollen allergen protein concentration and changes in graphene resistivity.

To verify the concentration dependence in Experimental Group 2, One-way ANOVA followed by Tukey’s HSD post-hoc test was performed on the resistivity change (Δρ) data of 1:5, 1:10, and 1:15 dilutions (n = 3 per group).

The independent variable was “allergen dilution ratio” (three levels: 1:5, 1:10, 1:15) and the dependent variable was Δρ, with 3 repeats per group. The Δρ data were as follows:

1:5 dilution (11.4158 × 10⁻⁶ Ω∙m, 13.1076 × 10⁻⁶ Ω·m, 10.8408 × 10⁻⁶ Ω·m);1:10 dilution (5.9525 × 10⁻⁶ Ω∙m, 5.8706 × 10⁻⁶ Ω·m, 5.0139 × 10⁻⁶ Ω·m);1:15 dilution (2.1673 × 10⁻⁶ Ω·m, 2.9246 × 10⁻⁶ Ω·m, 2.8583 × 10⁻⁶ Ω·m).

All analyses were conducted using GraphPad Prism 9.0 with a significance level (α) of 0.05.

Parametric tests like ANOVA require normality and variance homogeneity, which were verified prior to analysis:

Normality (Shapiro-Wilk test): All groups passed the normality test (1:5 group: W = 0.982, P = 0.956; 1:10 group: W = 0.975, P = 0.931; 1:15 group: W = 0.968, P = 0.905), with all P-values > 0.05 indicating data conformed to a normal distribution.Variance homogeneity (Levene test): The test produced F = 0.326 and P = 0.732 > 0.05, confirming that variances across the three groups were homogeneous. Both assumptions for ANOVA were satisfied.

One-way ANOVA was used to analyze overall differences in Δρ between groups:

Test hypotheses: Null hypothesis (H₀) assumed equal population means of Δρ across all dilutions (μ₁ = μ₂ = μ₃), while the alternative hypothesis (H₁) posited at least two means were unequal.Results: ANOVA revealed highly significant differences in Δρ among the three dilution groups (F (2, 6) = 128.76, P < 0.001), leading to the rejection of H₀.

To identify specific differing groups, Tukey’s HSD post-hoc test was performed:

Calculation process: The test statistic q was calculated as

Where SE is defined as

where:

q: standardized mean difference between two groups; x i ¯ : Sample mean of Δρ for the first group in pairwise comparison (1:5 dilution); x j : Sample mean of Δρ for the second group in pairwise comparison (1:10 dilution); x i ¯ − x j ¯ : Absolute mean difference between the two groups;SE: Standard Error of the mean difference;MSW: Mean Square Within groups, calculated from the ANOVA, representing the average variance within each dilution group; n i : Sample size of the first group in the pairwise comparison;nj: Sample size of the second group in the pairwise comparison.

Tukey’s HSD post-hoc test results are summarized below in Table 11:

Table 11. Tukey’s HSD post-hoc test results from above calculation (Experimental Group 2).

All pairwise comparisons showed significant differences, with q-values exceeding the critical threshold for α = 0.05, confirming that Δρ decreased significantly with the decrease of allergen concentration.

2.3.4. Electrochemical Verification

To verify the preliminary results from above experiments and graphene resistance measurements, we will conduct two groups of electrochemical experiments to verify the results.

1) Electrochemical experimental design

In this section, the cyclic voltammetry (CV) method of the CHI760E electrochemical workstation is used to measure chemically modified graphene with 3B4 Anti-Bet v1 antibody. The specific recognition ability, concentration measurement, and reaction time of the pollen protein and antibody-modified graphene electrode for pollen protein are explored.

Through changes in peak current (Ip), peak potential (Ep), and peak area (Ah) in CV detection [21], the specific binding process between pollen protein and the graphene surface can be tracked. Compared with the subsequent resistance detection, electrochemical detection is closer to the mechanistic explanation, verifying the rationality of the signal source of the graphene sensor.

A chemically modified graphene electrode was used here as the working electrode; set up a specific recognition group and the concentration recognition group:

The specific recognition group features PBS blank, 1:10 non-specific pollen solution (Amb A1), 1:10 specific pollen solution (Bet v1); The concentration recognition group includes PBS blank, Bet v1 pollen protein solution diluted 1:5 and 1:10 with PBS.

The detailed parameters for Electrochemical Cyclic Voltammetry (CV) are listed in Table 12.

Table 12. Electrochemical cyclic voltammetry (CV) experimental parameters.

2) Specific recognition verification

Table 13. Comparison of specific recognition peak areas.

Figure 13. CV curves of PBS, 1:10 Bet v1 solution, and 1:5 Bet v1 solution.

The CV test curves of blank PBS, specific pollen solution (Bet v1, 1:10 dilution), and non-specific pollen solution (Amb A1, 1:10 dilution) are shown in Table 13 and Figure 13.

It can be seen from the figure that the cyclic voltammetry curve of blank PBS basically shows no peak, and the integrated charge is close to zero, indicating that the background signal of the electrode system is extremely low in the absence of target pollen protein. In contrast, the specific pollen Bet v1 shows obvious redox peaks in multiple cycles, and its peak integrated charge (Ah) is stably on the order of 10⁻⁴ C, with both positive and negative peaks significantly higher than the blank control. Although the non-specific pollen Amb a 1 also shows a peak response, the peak area is generally on the order of 10⁻⁵ C, which is on average more than an order of magnitude lower than that of Bet v1. This order of magnitude is sufficient for the sensor to distinguish between specific and non-specific proteins, verifying the specific recognition ability of the modified graphene electrode.

3) Concentration trend verification

In further concentration verification experiments, the specific pollen protein solutions with different dilution ratios (Bet v1, 1:5 and 1:10) for cyclic voltammetry detection was selected and compared them with the blank PBS control.

As can be seen from Figure 14, the blank PBS curve has almost no peak, indicating that the background signal in the solution is extremely low. With the increase of pollen protein concentration, both the peak current (Ip) and peak area (Ah) of the cyclic voltammetry curve increase significantly: in the 1:10 diluted solution, the redox peak can already be clearly identified, while in the 1:5 solution, the peak current and integrated charge further increase significantly. Although the peak potential difference between the two concentrations fluctuates to a certain extent, it is generally within a reasonable range of 0.1 V.

Figure 14. CV curves of PBS, 1:10 Bet v1 solution, and 1:5 Bet v1 solution.

To further quantitatively analyze the relationship between concentration and peak area, we first smoothed the original curve using CHI760E Electrochemical Workstation software and determined the peak potential when the reaction occurs by combining the potential when the open-circuit potential (OCPT) tends to be stable. Subsequently, stable peaks at the corresponding potential were selected on each curve, and the area under the peak was calculated using the integration method.

where:

Q represents the charge quantity;I (E) represents the current at potential E.

The peak area data were obtained as shown in Table 14:

Table 14. Comparison of peak areas of Bet v1 solutions with different concentrations.

By linearly fitting the data points, the curve of peak area increasing with pollen concentration was obtained as shown in Figure 15:

Figure 15. Curves of peak areas (μC) vs Bet v1 pollen concentration (C).

The resulting equation is:

Q = k∙C + b

Q = 1 × 10 − 5∙C + 2 × 10 − 7 (R² = 0.9378)

where:

k represents the sensitivity of the graphene device;C represents the pollen protein concentration;b represents the background baseline charge quantity.

It should be emphasized here that due to the complexity of the electrochemical reaction process, the linear relationship fitted in this study is only used to reflect the overall trend, not a strict quantitative formula. The quantitative relationship for repeatable experiments will be derived from resistance measurement with a simpler reaction mechanism.

However, this result can still indirectly confirm that there is a positive correlation between the electrochemical response of the graphene sensor and the pollen protein concentration, providing solid support for the quantification of above resistance detection results.

4) Determination of detection time

In the binding process between antibodies and proteins, the time required for the reaction to stabilize is of great significance for the time control of graphene resistance measurement. This is very helpful for field deployment of automatic detection, because a better understanding of the reaction time allows the second resistance detection to be performed after the reaction is completed.

Cyclic voltammetry test results in Figure 16 and Table 15 show that the integrated charge (Ah) of the positive potential peak gradually increases with the number of scanning cycles and tends to stabilize after multiple cycles. In the first few cycles, the peak area grows rapidly (an increase of about 118% from cycle 1 to cycle 3), and then gradually slows down; the change in peak area from cycle 7 to cycle 9 is less than 13%, indicating that the system has basically stabilized. Considering that the scanning rate is 0.05 V/s, the potential ranges from −0.6 V to +0.6 V and back to −0.6 V corresponds to approximately 48 s/cycle, so the reaction time required to reach stability is about 6 - 7 minutes. Therefore, in the subsequent resistance measurement, to ensure that the antigen-antibody binding process fully occurs and considering the differences between graphene samples, 10 minutes can be used as the routine measurement time.

Figure 16. Enlarged view of the positive potential of S1_3B4_Bet v1_1:10.

Table 15. Comparison of positive potential peak areas of S1_3B4_Bet v1_1:10.

Figure 17. Curve of positive potential peak area vs. cycle number for S1_3B4_Bet v1_1:10.

Figure 17 shows that the peak area increases monotonically with the number of cycles and gradually approaches saturation, reflecting that the antigen-antibody reaction shifts to a stable stage.

This research focuses on a chemically modified graphene biosensor for pollen allergen detection, with experimental results verifying the specific recognition of pollen allergen types and quantitative response to pollen concentrations.

Traditional pollen detection methods suffer from critical shortcomings: manual counting is labor-intensive and low efficiency, while Skin Prick Tests (SPT) are subjective and carry anaphylactic risks. Emerging biosensors for allergens include optical (e.g., surface plasmon resonance, SPR), electrochemical (e.g., gold nanoparticle-based), and piezoelectric (e.g., quartz crystal microbalance, QCM) platforms. The graphene sensor proposed here offers distinct advantages:

Sensitivity

The four-wire resistance method detects resistivity changes down to 2.16 × 10⁻⁶ Ω⋅m (1:15 dilution), corresponding to a limit of detection (LOD) ~0.11 mg/mL for ragweed allergen. This is comparable to SPR sensors (LOD ~0.1 mg/mL) but outperforms QCM (LOD ~0.5 mg/mL) and conventional electrochemical sensors (LOD ~0.3 mg/mL). Graphene’s 2D structure and high carrier mobility enable femtomolar-level sensitivity (10⁻¹⁵ M) for biomolecules, suggesting further optimization (e.g., reduced graphene defects) could lower the LOD to clinically relevant concentrations.

Specificity

The specific-to-non-specific resistance change ratio (average 1500.88 Ω vs. 576.90 Ω, Experimental Group 1 is ~2.6:1, which is higher than gold nanoparticle-based electrochemical sensors (~1.8:1) and comparable to antibody-functionalized SPR (~3:1). This specificity arises from IgG antibody-antigen recognition and the p-type doping mechanism—non-specific proteins (e.g., birch Bet v1 binding to ragweed antibodies) do not induce Frontier Molecular Orbital (FMO) alignment or significant electron transfer, resulting in minimal resistance changes.

Portability and Cost

Unlike SPR (>$50 k) or electrochemical workstations (>$10 k), the graphene sensor uses a simplified four-wire circuit and can be integrated with microcontrollers (e.g., AD5940) for real-time monitoring at <$500 per device. This makes it suitable for field deployment (e.g., school-based pollen stations at Beijing No. 80 High School, as implemented in this study) and low-resource settings.

Experimental Group 2 revealed an exponential relationship between pollen concentration and resistivity change (y = 0.594e^−239.3x, R² > 0.98), which aligns with Langmuir adsorption kinetics for antibody-antigen binding. At low concentrations (1:15 dilution, binding is diffusion-limited, and small increases in concentration lead to proportional increases in the number of bound combinations, driving a linear rise in resistivity change. At higher concentrations (1:5 dilution), the graphene surface approaches saturation (AFM showed ~80% coverage of allergen-antibody combinations), and additional binding induces cooperative effects, accelerating electron transfer and leading to an exponential increase in resistivity.

This exponential relationship is scientifically meaningful for two reasons:

1) It confirms that the sensor’s signal is directly linked to specific binding (not non-specific adsorption), as non-specific interactions would produce a linear or flat response;

2) It enables quantitative detection over a wider concentration range, which covers the typical ambient pollen concentration during allergy seasons. The high R² values (>0.98) indicate the model’s stability and robustness, supporting future development of a calibration curve for clinical or environmental monitoring.

While the sensor demonstrates promising performance, there is still room to improve:

1) The graphene sensors were tested with only two allergens (Ragweed Amb A1, Birch Bet v1); expanding to other common allergens will require optimizing antibody modification for each target.

2) Environmental factors (e.g., humidity, temperature) may affect graphene’s conductivity; integrating a humidity/temperature sensor into the device will enable signal correction.

The potential future applications based on this research may include wearable sensors for real-time allergy risk assessment, networked environmental monitoring stations, water quality monitoring and food additives etc. The sensor’s low cost and portability also make it suitable for global health initiatives, particularly in regions where traditional pollen monitoring is unavailable.