Plasmonics co-integrated with silicon nitride photonics for high-sensitivity interferometric biosensing

1. Introduction

Integrated photonic sensors have been extensively explored in recent years due to their high sensitivity characteristics combined with mass production capabilities towards low-loss and label-free lab-on-chip diagnostics [1]. Photonic sensors rely on the detection of small changes in the refractive index of the surrounding medium by monitoring the associated phase change along the exposed waveguide. In this context, evanescent wave photonic sensors were demonstrated in various configurations including, photonic crystals [2–4], Silicon On Insulator (SOI) wire waveguides [5] and Si₃N₄ slot waveguides [6] showing bulk sensitivity values up to 900 nm per Refractive Index Unit (RIU) [4]. In all these approaches, however, sensitivity is limited by the partial exposure of the propagating optical mode to the liquid analyte [1], since only the evanescent wave interacts with the surrounding medium.

Plasmonic-based sensors have emerged as an attractive alternative sensing technology due to the enhanced sensitivity characteristics offered by the profound exposure of SPP modes to liquid analyte, resulting already in commercially available devices [7,8]. Different approaches of realizing plasmonic sensors have emerged over time, proposing schemes based on either Surface Plasmon Resonance (SPR) or interferometric layouts. SPR sensors are plasmonic thin-film based sensors, detecting changes in the refractive index occurring at the surface of the metal film [9]. Even though SPR sensors have already been successfully commercialized, they require bulky prism-based coupling configurations to excite the SPP mode, impeding their integration in a compact photonic chip. On the other hand, all-plasmonic interferometric sensors integrated on a chip, can mitigate this issue, allowing for highly compact devices [10–14]. However, high optical losses inevitably introduced in all-plasmonic interferometric layouts and the limited flexibility in optimizing the device performance restrict their deployment in practical and multifunctional sensor layouts, relying completely on plasmonic structures.

Selective co-integration of plasmonics with photonic waveguides has been more recently proposed as a promising alternative to leverage the unmatched benefits of plasmonics with the deployment of low-loss photonics for additional optical functionalities that may supplement or enhance the overall circuit performance. A variety of plasmonic waveguide configurations co-integrated with photonic waveguides, has been already demonstrated on SOI [15–22] and on Si₃N₄ waveguide platform [23–27], the vast majority of them relying, however, either on Dielectric-Loaded SPP (DLSPP) waveguides [15,16,22] or on oxide cladded plasmonic waveguides [21,23,27], inherently excluding sensing applications. Only a small number of open cladded plasmonic waveguides integrated in photonic platforms has been reported in the literature, utilizing noble metals [17,18,24,25] and more recently also CMOS compatible metal materials [28,29], enabling direct contact of propagating SPP modes with aqueous solutions. Successful approaches of plasmo-photonic sensors have been presented so far in either intensity-based sensor configurations [30] or interferometric based sensor layouts [17,18,31]. The first approach suggests Short-Range SPPs with bulk sensitivity equal to 1365 dB/RIU while on the second approach hybrid plasmonic slot waveguides [17,31] or hollow hybrid plasmo-photonic waveguides [18] are utilized, yielding bulk refractive index sensitivities up to 1060 nm/RIU [17].

In this paper, we demonstrate an open cladded gold plasmonic stripe integrated as the sensing arm of a Si₃N₄-based MZI configuration for sensing applications. More specifically, a butt-coupled interface between Si₃N₄ waveguides and a gold stripe is adopted in one branch to serve as the sensing transducer by detecting local changes in the refractive index, resulting in an interferometer resonance shift at the sensor output. The proposed device takes advantage of the low-loss Si₃N₄ platform to deploy a MZI-based variable optical attenuation stage followed by a thermo-optic phase shifter in the reference branch in order to optimize the sensor performance. The variable optical attenuator (VOA) is used to balance optimally the losses of the two MZI branches and therefore maximize interference at the sensor output achieving spectral extinction ratio (ER) values of more than 35 dB. The additional phase shifter was added as the means to tune the MZI resonance in the spectral window of operation. The proposed sensor was experimentally evaluated revealing a bulk refractive index sensitivity value of 1930 nm/RIU, which is, to the best of our knowledge, the highest reported sensitivity so far among all plasmo-photonic sensors and a Figure Of Merit (FOM) value of 161 at 1565 nm.

2. Sensor layout and fabrication

The integrated sensor involves a MZI where a “plasmo-photonic” waveguide branch is utilized to detect changes in the refractive index of test analytes. More specifically, when a refractive index change occurs on the surrounding medium of the sensing branch, it will introduce a phase change, experienced by the propagating SPP mode, resulting, eventually, in a resonance spectral shift at the interferometer output. Figure 1(a) depicts a conceptual schematic of the proposed plasmo-photonic MZI sensor, illustrating also the fiber-to-chip coupling that relied on a vertical out-of-plane coupling scheme based on Si₃N₄ Grating Couplers (GCs) for Transverse Magnetic (TM) polarization [32].

 

Fig. 1 (a) Conceptual schematic of the plasmo-photonic sensor, showing the fiber-to-chip coupling. (b) Schematic of the plasmonic sensing area. (c) Cross-sectional dimensions for the photonic and the plasmonic waveguides across the direction of propagation.

Download Full Size | PPT Slide | PDF

The photonic waveguide platform was based on a 360 nm × 800 nm Si₃N₄ core lying on top of a 2.2 μm thick SiO₂ substrate layer and cladded with 600 nm of Low Temperature Oxide (LTO) (Fig. 1(c) I). The gold stripe is recessed in a cavity formed in between the photonic tapers Fig. 1(c) II. In addition, the gold stripe resides on a Ti layer located between the gold film and the SiO₂ layer. The Ti layer serves as both the adhesion layer of gold in the cavity and in parallel to avoid coupling of top surface plasmons into radiation modes in the buried oxide layer [22] (Fig. 1(c) III). The plasmonic stripe is deposited in the cavity 500 nm away from the tapered photonic waveguide facets (longitudinal offset) and a 400 nm below the photonic waveguide level optimizing photonic to plasmonic coupling efficiency as described in [24]. The reference branch comprises a thermo-optic heater-based phase shifter for tuning the sensor’s resonance wavelengths and a VOA for balancing optical losses. The phase shifter consists of two 100 nm × 2 μm titanium-based resistors, placed on top of the photonic waveguide, in a double-heater configuration ((Fig. 1(c) IV). We deployed two parallel Ti-based metallic wires on top of the LTO cladding and symmetrically positioned away from the Si₃N₄ waveguide in order to minimize the overlap of the photonic TM mode (exposed to cladding), with the titanium metal wires while maintaining the thermo-optic effect. The VOA composes of a MZI, with a thermo-optic heater-based phase shifter in one branch, in order to control the attenuation of optical power with applied voltage. This attenuation stage is utilized to counter-balance, the losses introduced in the plasmonic waveguide of the sensing arm, eventually maximizing field interference at the MZI output. Moreover, two low-loss 3-dB Multi-Mode Interference (MMI) Y-junctions were utilized to allow for equal power splitting ratios between the MZI branches.

Figure 2(a) illustrates a highlight of the mask layout presenting the proposed plasmo-photonic sensor layout, employing on the reference arm a thermo-optic phase shifter (PS1) and the VOA-MZI structure with a phase shifter in its upper arm (PS2). Figure 2(b) demonstrates a photo of the fabricated sensor chip. The photonic waveguides have been fabricated using 150 mm silicon wafers where a 360 nm thick Si₃N₄ layer was deposited on top of a 2.2 μm thick thermal oxide in a Low-Pressure-Chemical-Vapor-Deposition (LPCVD) process.

 

Fig. 2 (a) Mask layout of the plasmo-photonic sensor. (b) Photo of the fabricated sensor chip illustrating the fluidic tubes glued on top of the chip. (c) Cross-sectional SEM image of the photonic waveguide. (d) Angled SEM image of the etched cavity (red dashed line) and the Si₃N₄ tapers (white dashed line) before the plasmonic waveguide deposition. (e) SEM top-view image of the gold plasmonic waveguide recessed in the cavity.

Download Full Size | PPT Slide | PDF

Marker layer and waveguides have been defined by optical projection lithography using an i-line stepper tool. Reactive-Ion-Etching (RIE) with CHF₃ and He chemistry has been employed for the structure transfer. After the structure transfer, a 600 nm thick LPCVD SiO₂ was deposited to form the LTO cladding of the photonic waveguide and annealed at 1000 °C for several hours.The annealing process improved the optical properties of the oxide being associated with the induced optical losses due its hygroscopic properties. A cross section view of the fabricated photonic waveguide is depicted in Fig. 2(c).

The plasmonic cavity has been defined also by the same stepper tool and etched via RIE through LTO, Si₃N₄ and 400 nm thermal oxide. The resulting cavity depth was 1.36 μm. Figure 2(d) shows a scanning-electron-microscope (SEM) image of the proposed cavity (red dashed line) before the deposition of the gold plasmonic stripe while the white dashed lines highlight the photonic linear tapers. In the final fabrication stage, the metallic plasmonic stripes were deposited into the cavities by a lift-off process using e-beam lithography, thermal evaporation of gold and lift-off dissolution. The target thickness for gold was 100 nm. A very thin (3 nm) titanium layer was deposited by e-gun evaporation prior to gold for an improved adhesion of gold on the sample. A SEM top-view image of fabricated 70 μm long gold plasmonic waveguide, recessed in the cavity, is shown in Fig. 2(e). Silicone fluidic vessels were also glued on top of the plasmonic sensing waveguides as the means to confine the test liquid in this region alone, shown in more detail in Fig. 2(b).

3. Experimental results

Prior proceeding with the evaluation of the complete sensor, we tested the functionality of a stand-alone VOA structure to validate its capability for dynamically controlling the VOA-induced attenuation level. The VOA comprises a 1x1 thermo-optic MZI structure similar to the one employed on the sensor reference arm and its mask layout is illustrated in Fig. 3(a). Electrical power was applied to the thermo-optic phase shifter and the respective optical attenuation on the output of the MZI was recorded.

 

Fig. 3 (a) Mask layout of the 1x1 MZI structure. (b) Measured optical attenuation with varying electrical power on the thermo-optic phase shifter of the MZI.

Download Full Size | PPT Slide | PDF

Figure 3(b) depicts the resulted induced loss versus applied electrical power, revealing a curve that is close to the typical interferometric response of a thermo-optic MZI and yields a maximum attenuation of 33.257 dB. Next, we evaluated the functionality of each individual element of the fabricated sensor. To do so, broadband Fiber-to-Fiber (FtF) optical measurements were carried out by sweeping the wavelength of a Tunable Laser Source (TLS) from 1500 to 1580 nm, while a polarization controller and a Polarization Maintaining (PM) fiber were used at the input of the chip to inject TM polarized light into the chip. Light after propagating through the sensor device was collected via an output Single Mode (SM) fiber and fed to an optical power meter. In order to evaluate the attenuation, inserted by the VOA in the reference branch, the titanium heaters were connected to an external voltage source via aluminum-based electrical pads. Figure 4(a) shows the sensor spectral response at various electrical power values, showing that different ER values are obtained for different levels of optical attenuation corresponding to different values of electrical power applied to the VOA heaters. More specifically, when applying 0, 90, 180, 280, 504 mW of electrical power to PS2 we obtain the corresponding curves with ER equal to 24.5, 7.38, 4.75, 4.38, 11.75 dB, respectively. The maximum extinction ratio is 24.3 dB when water covers the sensing branch, achieved at 0 mW. Next, the effect of the phase shifter is demonstrated by the graph of Fig. 4(b). In this case, electrical power was applied to the heaters of the phase shifter to tune the sensor resonance wavelengths by retaining a constant optical attenuation level at the VOA.

 

Fig. 4 (a) Measured sensor spectral response with varying electrical power on the thermo-optic VOA. (b) MZI blue-shift for two power dissipation levels on the thermo-optic phase shifter.

Download Full Size | PPT Slide | PDF

As can be seen in Fig. 4(b), the resonant wavelengths are blue-shifted as the electrical power applied to the phase shifter increases from 0 mW to 112 mW, validating the tunability of the resonant dips. The total FtF losses of the employed sensor device were found to be 39.8 dB at 1550 nm, incorporating insertion losses of two gratings (2 × 14.5 = 29 dB), the Si₃N₄ propagation loss of the 0.511 cm long sensor (0.55 × 0.511 = 0.28 dB), splitting loss of 4 MMI couplers (0.2 × 4 = 0.8 dB), insertion losses from two photonic tapers (2 × 0.2 = 0.4), insertion losses from two Si₃N₄-to-Au interfaces (2 × (2.3 ± 0.3) = 4.6 ± 0.6 dB) [24] and the propagation losses of the 70 μm Au plasmonic waveguide (70 × (0.058 ± 0. 0.0036) = 4.06 ± 0.252 dB) [24]. Consequently, we conclude that the FtF losses of our plasmo-photonic sensor are dominated by the I/O coupling loss (14 dB/grating), while the losses coming from the plasmonic stripe are retained under 10 dB. The excessive I/O loss can be attributed to fabrication deviations from the targeted grating coupler’s design parameters, however the I/O coupling loss can be significantly improved either by improving the fabrication of the gratings so as to match closer the targeted performance reported in [32] or by employing an edge coupling scheme on Si₃N₄ [34], reducing the I/O coupling loss down to 2 dB/taper.

Sensitivity measurements were accomplished by infiltrating 4 different buffer solutions with refractive indices ranging between 1.3327 and 1.3378, into the fluidic vessel (sensing area) and the respective sensor spectral response was recorded. Figure 5(a) shows the MZI spectra when the plasmonic stripe is infiltrated with distilled water (black curve), two aqueous solutions with different concentrations of Phosphate-Buffered Saline (PBS) (blue curve, red curve) and NaCl solution (green curve), clearly revealing that the resonances are shifting to longer wavelengths with increasing refractive index of the buffer solution. The extinction ratio of the shifted resonances varies with the buffer solution utilized. This is mainly attributed to the change of the SPP propagation losses when different buffer solutions are used, which therefore changes the power balancing conditions of the MZI. Figure 5(b) depicts the obtained spectral response when PBS solution is injected in the fluidic vessel, revealing an FSR ranging from 26.6 nm at (1512 – 1538.6) nm wavelength range to 24.8 nm at (1538.6 – 1563.4) nm wavelength range, and a maximum extinction ratio equal to 37 dB at 1571.6 nm. Figures 6(a)-6(c) illustrate in more detail the red spectral shift for the 3 different interferometer resonances around 1510 nm (Fig. 6(a)), 1540nm (Fig. 6(b)) and 1565nm (Fig. 6(c)), as the refractive index value of the buffer solution increases. The differences in refractive index between distilled water (1.3327) and the different buffer solutions were calculated, indicating values equal to 0.0003, 0.002 and 0.0051 for the case of PBS1, NaCl, PBS2 solutions respectively.

 

Fig. 5 (a). Sensor spectral response obtained when injecting 4 different buffer solutions. (b) Enlarged plots of the sensor spectral response, revealing the high ER value of 37 dB.

Download Full Size | PPT Slide | PDF

 

Fig. 6 (a)-(c) Enlarged plots of the sensor spectral response experimentally obtained when injecting buffer solutions with different refractive indices revealing a red shift of the resonance wavelength. (d)-(f) Resonance shift versus refractive index of the buffer solutions, illustrating linear fitting plots for each of the three dips, for calculating the sensitivity of the sensor.

Download Full Size | PPT Slide | PDF

The corresponding wavelength shifts equal to 2.1, 2.4 and 6.8 nm for the resonant dips around 1515 nm, 1.9, 4.6, 8.4 nm for the resonant dips around 1545 nm and 1.3, 5.4, 9.2 nm for the resonant dips around 1565 nm. Consequently, bulk sensitivity values were calculated for each of the three resonant dips by applying least-squares linear fitting on the resonant wavelength shift with increasing refractive index change, as shown in Figs. 6(d)-6(e). Bulk sensitivity increases with longer wavelength resonances, probably owing to the waveguide dispersion characteristics around the three different resonant wavelength spectral regions. Figure 7(a) shows the achieved bulk sensitivity as a function of wavelength, ranging from 1332 nm/RIU around 1515 nm to 1930 nm/RIU around 1565 nm.

 

Fig. 7 Bulk refractive index sensitivity as a function of wavelength.

Download Full Size | PPT Slide | PDF

In order to further evaluate the performance of the proposed plasmo-photonic sensor and demonstrate the usefulness of the VOA, we calculated the Figure Of Merit (FOM) value for 3 different states of the VOA at 1565 nm, as illustrated in Fig. 4(a), according to the following formula [31]:

4. Conclusion

We have successfully demonstrated a novel liquid refractive index sensor based on an integrated plasmo-photonic MZI configuration. Our approach exploits the high sensitivity characteristics of a liquid-loaded gold plasmonic stripe integrated in Si₃N₄ photonic platform to achieve the record-high bulk sensitivity value, among all plasmo-photonic sensors, of 1930 nm/RIU. The Si₃N₄ circuitry allows for multiple low-loss on chip functionalities, such as a VOA stage in the reference branch leading to extinction ratio values of the MZI interferometer of more than 35 dB. The proposed sensor is a tangible example of the benefits offered when the high-sensitivity capabilities of plasmonics get combined with the low-loss and low-cost LPCVD-based Si₃N₄ platform in compact, low-cost and ultra-sensitive interferometric sensing devices [33].