Carbohydrate-protein interactions characterized by dual polarization hybrid plasmonic waveguide

2.1 Schematic of DPHP waveguide

Figure 1A presents the three-dimensional (3D) schematic structure of the proposed DPHP waveguide, consisting of a silicon-on-insulator (SOI) rib and a metal (Ag) cladding. The region between the Si core and the Ag layer is filled with a tested analyte. In this way, double nano-slots are formed at both sides of the Si core as well as another slot is embedded on the top of the Si core. The cross-section view with geometric parameters is presented in Figure 1B. The widths of the Si core and the lateral nano-slot are denoted as w_si and w_slot, respectively. The height of the Si ridge and the upper nano-slot are defined as h_si and h_slot, respectively. The optimized values of w_slot and h_slot are setting as 40 nm and 60 nm, respectively, giving rise to a strong filed confinement and a low propagation loss. In Figure 1C, a field profile obtained by simulation reveals that the present configuration can support both the TE and TM modes. The values of the geometric parameters were assigned: w_si equals 550 nm; h_si equals 320 nm; h_sio2 (substrate) is 3 μm high, and the length of the waveguide (l) is 25 μm. The thickness of the Ag cladding is set at 3 μm so as to neglect the influence of the simulation boundaries. This structure is manipulated compatibly with the conventional SOI optical as well. In the simulation work, the refractive indices [27] of Si and SiO₂ are set as 3.5046 and 1.447, respectively, at a central wavelength of 1.3 μm. The wavelength selection [7] is based on a tradeoff between the absorption loss caused by the aqueous solution and the broadband light sources. The water solution optical absorption is smaller at a wavelength of 1.3 μm, compared with the 1.55-μm wavelength commonly used for operation. The permittivity of silver is achieved through the Durde model [28]:

Figure 1:

(A) Schematic of DPHP waveguide. (B) Cross-section view of the DPHP waveguide filled by test liquid. The values of the geometric parameters are w_slot=40 nm, h_slot=60 nm, w_si=550 nm, h_si=320 nm, h_Ag=3 μm, h_sio2=3 μm, and l=25 μm. (C) Field profile of the DPHP waveguide.

where ε_∞=3.1, ω_p=140×10¹⁴ rad/s, and γ=0.31×10¹⁴ rad/s.

2.2 Theory of DPHP waveguide

For most waveguide systems, the evanescent wave decays exponentially with the decay length in the order of 0.1–1 μm [20]; thus, it only senses the changes taking place on the surface of the waveguide due to the higher intensity of the evanescent field in this particular region. Therefore, the shifts in the bulk solution is hardly considered, whereas the bulk index has an exact influence on the sensor response. With regard to the proposed structure, the incident light goes through the waveguide and excites the surface plasmon in the slots when the phase velocities of the optical mode and that of the surface plasmon match. The field is highly confined in the slots and sufficiently penetrates into the solution. As a result, the bulk refractive index can be measured. Evidently, the slots are able to store electromagnetic energy resulting in subwavelength optical guiding with lower propagation loss. The dielectric discontinuity at the Si core solution interface produces a polarization charge [29], which interacts with the plasma oscillations of the metal-solution interface. The gap region can be equated with an effective optical capacitance [29] resulting in a strong field confinement and a lower propagation loss. The propagation constant of the surface plasmon depends strongly on the wavelength, in comparison with the optical modes of the traditional dielectric waveguide. The phase-matching condition between an optical mode of the dielectric waveguide and a surface plasmon supported by metal (Ag) cladding might be fulfilled within a narrow spectral band. Therefore, when a broadband light (1.285 μm≤λ≤1.315 μm in this case) is applied to the waveguide, the spectrum presents a narrow dip with optical energy conversion into the surface plasmon [30]. The position of the resonance dip in the output spectrum strongly relies on the refractive index of the sensed medium. Any changes in the sensed medium will alter the resonance condition, which results in a shift in the resonance dip. Thus, we can adopt wavelength interrogation to interpret the molecular interaction.

2.3 Carbohydrate-protein interactions based on DPHP waveguide

Herein, HepV-heparin interaction was emulated based on the FEM and carried out with COMSOL. Generally, the HepV-heparin binding event can be simplified into four distinct steps [25]: (1) Tris buffer flowing over the biotin-functionalized waveguide surface for initial calibration, (2) streptavidin immobilized on the surface, (3) the capture of biotinylated heparin, (4) introducing recombinant HepV fragment to recognize the biotinylated heparin. During the biological event, the thickness a and the refractive index n_l of the adlayer with a known numerical value were obtained from the typical literature data [25]. Besides, the bulk refractive indices, in this case, were calculated according to the empirical equation [25] expressed as n_l=0.52*(n_p−n_s)+n_s, where n_p is the refractive index of pure protein (1.465) [31], [32], [33] or solvated carbohydrate (1.45) [34]. All the data, including n_l, n_s and a, are presented in Table 1. The effective mode indices (n_{eff_TE} and n_{eff_TM}) were measured for a wavelength spanning 1.285 μm≤λ≤1.315 μm at all stages. The fitting curves of the n_{eff_TE} and n_{eff_TM} versus the wavelength λ are plotted in Figure 2. One can see clearly the process of the HepV-heparin binding event in the whole spectrum range. The shifts in the resonance wavelength depending on the effective mode indices of the waveguide are significantly different. Thus, we can take the wavelength interrogation to interpret the molecular interaction.

Figure 2:

Simulation results of HepV-heparin binding event at a wavelength spanning 1.285 μm≤λ≤1.315 μm for the TE mode (A) and the TM mode (B), respectively. A plot of wavelength versus effective mode index provides a representation of each step.

2.4 Property of DPHP waveguide

In general, sensitivity and propagation loss are considered as significant criteria to evaluate the performance of a biosensor. It is necessary to validate these two physical quantities of the proposed DPHP waveguide. The waveguide sensor sensitivities respond to (1) the changes in the bulk refractive index ∆n_s, (2) the shifts in the biolayer index Δn_l, and (3) the adlayer thickness variation Δa. In the molecular interaction, all the effects occur, and consequently, the resulting effective mode index change is:

where ∂nneff_i∂ns, ∂nneff_i∂nl, and ∂nneff_i∂a represent, respectively, the shifts in the bulk index, the biolayer index, and the thickness contributing to the effective mode indices. For the wavelength shift detection, we investigate the transmission spectrum on the resonance wavelength spanning 1.285 μm≤λ≤1.315 μm. The data are plotted in Figure 3. The position resonance dip is dependent on the effective mode index of the waveguide as well as on the polarization. With the process of the molecular event, the resonance wavelength for the TE mode moves to short-wave direction, while for the TM mode, the resonance wavelength shifts toward the longer wavelength. Various numerical values of the blue-shifted and red-shifted wavelengths arise from different influences of the effective mode index, which are also in agreement with the results in Figure 2. For TE mode, compared with the Tris buffer calibration, the resonance wavelength caused by streptavidin immobilization is blue shifted to 6 nm, while the resonance wavelengths induced by heparin capture and HepV binding are blue shifted to 2 nm and 4 nm, respectively. For the TM mode, a redshift of 8 nm with streptavidin immobilization and that of 12.5 nm and 10 nm with heparin capture and HepV binding, respectively, are presented.

Figure 3:

The dependence of resonance wavelength on transmissions during the whole process for both TE mode (A) and TM mode (B).

Afterward, the spectral interrogation sensitivities, caused by changes in bulk refractive index ∆n_s for dual polarizations can be defined as follows:

where ∂λ∂nneff_i is the resonance wavelength shift caused by the effective mode index of the DPHP waveguide, which was measured at each step (Figure 2), and ∂nneff_i∂ns represents the effects of the buffer solution on the effective mode index. Similarly, the spectral interrogation sensitivities, due to the adlayer index variation Δn_l can be interpreted as below:

where ∂nneff_i∂nl is the change in the effective mode index resulting from the adlayer indices. Apart from Δn_s and Δn_l, the spectral interrogation sensitivities caused by the adlayer thickness can be expressed:

where ∂nneff_i∂a represents the shift in the adlayer thickness contributing to the effective mode index of the waveguide. Therefore, the resonant wavelength shift in complete formula is:

where Δλ_{DPHP_i} is the shift in resonance wavelength, and Δn_s, Δn_l, and Δa are shifts in the bulk index, adlayer index, and thickness in each step (excluding step 1) compared with the calibration step.

In order to investigate the propagation loss for the hybrid plasmonic waveguide, the optical confinement factor Γ, defined previously [35], which was power confined in a particular area divided by the total power, was introduced. It has been shown [35] that Γ is not only influenced by the loss-less photonics mode but also by the lossy plasmonic mode. Thus, we can improve Γ_layer (power confined in the layer compared with total power) by tuning the geometrical properties of the structure to gain lower loss. The optimized size of the waveguide is set as aforementioned: w_slot=40 nm, h_slot=60 nm, w_si=550 nm, h_si=320 nm, h_Ag=3 μm, h_sio2=3 μm, and l=25 μm. To calculate the propagation loss, we derive the loss coefficient α (in dB/μm) in the following equation: α=−lgPoPi/2lge, with P_o equals the power outflow time average (w) and Pi equals the power inflow time average (w) and depicting the loss variation at a wavelength spanning at 1.285 μm≤λ≤1.315 μm in Figure 4. The propagation loss was measured for the optimized dimension of the DPHP waveguide with water solution for simplicity. It is presented that the optimum value for the propagation loss achieves as low as the order of 10⁻⁴.

Figure 4:

Dependence of propagation loss on the wavelength varying from 1.285 μm to 1.315 μm for the optimized waveguide dimension for dual polarization. The value for loss can reach as low as the order of 10⁻⁴.

2.5 Characterization of heparin-Hep V interaction

To extract the biolayer thickness and index, we rewrite equation (6) with fixed wavelengths at resonance dips (λ=1.291 μm, λ=1.295 μm, λ=1.293 μm for the TE mode and λ=1.303 μm, λ=1.3075 μm, λ=1.305 μm for the TM mode):

where Δλ_{DPHP_i} (i=TE,TM) are shifts of resonance wavelengths at all stages compared with the Tris buffer calibration, and the numerical values can be achieved via Figure 3.

The ∂λ_i∂ns, ∂λ_i∂nl, and ∂λ_i∂a (i=TE,TM) are wavelength shifts induced by the bulk index, adlayer index, and thickness, respectively. To calculate ∂λ_i∂ns, for example, we can divide the term into ∂λ∂nneff_i, which can be obtained through Figure 2, and ∂nneff_i∂ns, which can be measured at resonance dips at each step (Figure 5) and subsequently doing the math. With the same method, ∂λ_i∂nl and ∂λ_i∂a can be achieved when integrated with Figures 2, 6 and 7 . For the relationship between the bulk index and the adlayer index, in this case, n_l=0.52*(n_p−n_s)+n_s, and the expression simplifies to:

Figure 5:

The variation of transmission arising from a different bulk refractive index (1.33≤n_s≤1.37 with step 0.01) in the molecular binding event. (A) λ=1.291 μm for the TE mode, (B) λ=1.303 μm for the TM mode.

Figure 6:

The variation of transmission caused by biolayer index changing (1.4≤n_l≤1.41 with step 0.01) during molecular interaction. (A) λ=1.295 μm for the TE mode, (B) λ=1.3075 μm for the TM mode.

Figure 7:

The variation of transmission induced by the shift in adlayer thickness (5 nm ≤a≤6 nm with step 0.2 nm) during the molecular interaction. (A) λ=1.293 μm for the TE mode, (B) λ=1.305 μm for the TM mode.

It is noteworthy that to simplify the numerical calculation, we substitute n_s with n_l via the empirical formula [25] n_l=0.52*(n_p−n_s)+n_s and, thus, combine ∂λ_TE∂ns with ∂λ_TE∂nl,and ∂λ_TM∂ns with ∂λ_TM∂nl, which were subsequently denoted as ∂λ_TE∂nl∗ and ∂λ_TM∂nl∗ including both the effects of n_s and n_l. The calculated thickness a and the refractive index n_l via equation (8) share similar results compared with the numerical values measured by DPI [25].

Given the values of biolayer thickness and refractive index, the surface mass density during the binding event can be calculated through the De Feijter’s formula [36]: ρ_l=(n_l−n_s)/(dn/dc), where dn/dc is the refractive index increment (in cm³·g⁻¹), which in the case of protein and carbohydrate are 0.186 cm³·g⁻¹ and 0.142 cm³·g⁻¹, respectively. The densities of the adlayer dedicated by streptavidin, biotinylated heparin, as well as HepV, are 0.357 g·cm⁻³, 0.322 g·cm⁻³, and 0.336 g·cm⁻³, respectively. Integrating with adlayer thickness and density, we can conclude that a thick sparse layer (a=5.46 nm, ρ_l=0.322 g·cm⁻³) is formed after heparin capture, and a thin dense layer (a=5.33 nm, ρ_l=0.336 g·cm⁻³) is formed as HepV is bound. This phenomenon can be explained by HepV undergoing conformational change on binding to heparin and a loss of streptavidin from the waveguide surface during HepV injection, thereby, a decrease in adlayer thickness and density, compared with that of streptavidin introduction.

2.6 DPHP waveguide fabrication

We have envisaged a CMOS-compatible fabrication process of the designed device. The process can start from a commercial SOI wafer. Alternatively the plasma-enhanced chemical vapor deposition (PECVD) technology is used to deposit the SiO₂ and α-Si thin film on the Si substrate. The photoresist thin film is formed on the Si film, and the waveguide patterns are constructed using an E-beam lithography so as to gain high resolution. Then, an extreme thin aluminum (Al) film, with a 40-nm thickness on the sidewall and a 60-nm thickness on the top of the Si rib, was deposited by the sputtering process. This step could be followed by the deposition of a metal (Ag) layer with a 3-μm thickness on the sidewall and at the top of the Al film by the sputtering process. Afterward, the Al film is removed by selectively etching to form the vertical and horizontal slots.