Fast, accurate, point-of-care COVID-19 pandemic diagnosis enabled through advanced lab-on-chip optical biosensors: Opportunities and challenges

The sudden rise of the worldwide severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic in early 2020 has called into drastic action measures to perform instant detection and reduce the rate of spread. Common clinical and nonclinical diagnostic testing methods have been partially effective in satisfying the increasing demand for fast detection point-of-care (POC) methods to slow down further spread. However, accurate point-of-risk diagnosis of this emerging viral infection is paramount as the need for simultaneous standard operating procedures and symptom management of SARS-CoV-2 will be the norm for years to come. A sensitive, cost-effective biosensor with mass production capability is crucial until a universal vaccination becomes available. Optical biosensors can provide a noninvasive, extremely sensitive rapid detection platform with sensitivity down to ∼67 fg/ml (1 fM) concentration in a few minutes. These biosensors can be manufactured on a mass scale (millions) to detect the COVID-19 viral load in nasal, saliva, urine, and serological samples, even if the infected person is asymptotic. Methods investigated here are the most advanced available platforms for biosensing optical devices that have resulted from the integration of state-of-the-art designs and materials. These approaches include, but are not limited to, integrated optical devices, plasmonic resonance, and emerging nanomaterial biosensors. The lab-on-chip platforms examined here are suitable not only for SARS-CoV-2 spike protein detection but also for other contagious virions such as influenza and Middle East respiratory syndrome (MERS).


I. INTRODUCTION
A. Importance of highly sensitive point of care detection The Corona virus disease 2019 (COVID- 19) pandemic and its rapid growth rate has driven an unprecedented worldwide demand for measures to mitigate its fast spread rate [1][2][3] .
Adding mandatory large-scale policies like social distancing and extreme antiviral disinfection measures and protocols in coping with infected patients, it becomes paramount to have detection and curing system developed in a catastrophic crisis state of which very few mankind alive have experienced before. It is a commonly held view that pressures of war have stimulated advances in engineering, science, and medicine. Therefore, the new invisible battle against SARS-CoV-2 virus infection can stimulate major breakthroughs in the development of diagnosis and treatment systems. SARS-CoV-2 highly contagious infection is hard to detect as patients can be present with clinically inapparent symptoms including fever, cough, or shortness of breath 4 . The worldwide morbidity and mortality of SARS-CoV-2 plus no available vaccine or guaranteed treatments on the horizon as of mid-2020 bolds the necessity for researchers to probe various medical interventions. Immediate cost-effective point-of-risk measures like identification, diagnosis and isolation of the infected individual is still regarded as the single best viable solution to slow down this pneumonia pandemic.
B. Structure of the SARS-CoV-2 SARS-CoV is an enveloped, single stranded RNA virus, that exists in humans and animals, and is mainly transmitted through aerosols and nearby interpersonal contacts 5,6 . Once the virus enters the body, it sticks to primary target cells which provides plenty of virus receptors, the angiotensin-converting enzyme (ACE2) 5,7 . Its genome RNA infusion into the cell results in the formation of protein building blocks consist of spike, envelope, membrane, nucleocapsid, and proteins [8][9][10] . Thus, human SARS-CoVs relies heavily on ACE2 for infusion into the target cell for which S glycoprotein trimeric spikes on the surface mediates the entrance into the host cell 11 . The S glycoprotein of SARS-CoV is therefore the main target for neutralizing antibodies (nAbs) 12 . Similar SARS-CoV and SARS-CoV-2 amino acid identity in their S proteins makes them prone to have analogous Immunogenic surfaces on these antigens 13 . Coronaviruses demonstrate a complex pattern for receptor recognition 7,14,15 (Figure 1.c). The attempts to block the infusion of virus has been carried out through targeting mainly spike protein of SARS-CoV-2 and the receptor binding domain (RBD). Antibodies developed specifically for these regions can expand the potency and power and chance of success against the infusion of SARS-CoV-2 in the host cell. Once the virus enters the body through the cells, it replicates and virions are then set free to infect new target cells 16,17 . SARS Infectious viral particles can be found in respiratory secretions, urine and sweat. SARS-CoV infection harms lung tissues resulting in pneumonia with rapid respiratory deterioration and failure and in almost %5 of cases, death 18,19 .
Development of effective vaccine against SARS-CoV-2 can be effectively applied through S protein and especially the receptor binding domain (RBD) as they induce highly potent neutralizing antibody to block virus binding and its membrane infusion or forming immunity protective layer against viral infection 5 .

C. Detection mechanisms for SARS-CoV-2 viral infection
The most standard procedure for identifying pathogens like SARS-CoV-2 relies on real-time reverse transcription-polymerase chain reaction (RT-PCR) in which the virus ribonucleic acid (RNA) molecules ( Fig. 1) go through a time consuming labeling procedure known as reverse transcription (RT) 20 . For patients who present late with a viral load below the detection limit of RT-PCR assays, serological diagnosis with less sensitivity is also used.
The aforementioned process and similar clinical diagnosis requires advanced laboratories, equipment and expertise hard to be found in under-developed remote areas that are more prone to the outbreak [20][21][22][23][24] .
The test is done for a qualitative analysis of nucleic acid from the SARS-CoV-2 gathered from people who meet SARS-CoV-2 virus clinical infection signs and symptoms (Fig 1.b). It has also been reported that SARS-CoV-2 can be clinically detected from saliva, blood, and urine samples in 30 to 45 minutes. Considering the high cost and time-consuming nature of clinical diagnosis procedure like RT-PCR (minimum 3 hrs 25 ) and test kit requiring samples generated by urine, saliva, blood and nasopharyngeal swabs (minimum 30 minutes 9-11 ), the need to develop a fast-accurate detection method for SARS-CoV-2 is better recognized. Biosensors can provide the next best alternative reliable solution to clinical diagnosis with much faster real-time detection without compromising sensitivity and accuracy 29,30 . Optical biosensors are particularly likely to become the future COVID-19 diagnostic tools [31][32][33][34][35] . By exploiting the strong light-matter interactions, one can create an ultra-sensitive label-free real-time detection platform for novel SARS-COV-2. Primarily, an optical biosensor translates the capture of the target analyte in a measurable alteration of a light property, such as refractive index (RI), intensity or resonance shift, through different methods such as resonators and interferometers ( Fig. 2).
However, due to time-sensitivity and shocking nature of the COVID-19 pandemic, early efforts for detection have been mainly based on how existing systems can be integrated together competently to outperform the existing processes in respect to sensitivity and timeconsumption. It has been demonstrated by Chen et al. and Alam et al. 36 Combination of plasmonic photothermal effect and localized surface plasmon resonance sensing transduction has also been proposed as promising for COVID-19 diagnosis 29 .

D. Principle of lab-on-a-chip optical biosensors
To spot and monitor the real-time binding of small numbers of biomolecules such as proteins, biosensors with ultra-high sensitivity are required. Optical transducers have been extensively researched, commercialized and deployed in hospitals. Label-free optical biosensing can provide sensitive and durable point-of-care testing (POCT) device which is imperative for SARS-CoV-2 epidemic containment as it also can be easily operated at the point-of-risk by individuals without specialty training 40 (Fig. 2). On top of that, label-free optical biosensors have exhibited a strong conceivable potential to grow expeditiously in healthcare and biomedical fields as they provide a condensed accurate analytical tool to promote mass-scale screening of a broad range of samples through different parameters 35,41,42 . Optical biosensing does work in different physical transduction principles such as interferometers, resonators and plasmonic 43 and has been investigated to monitor many viruses with a good accuracy in different studies 31,44,45 .

E. Surface functionalization strategies
The difficulty of measuring physical features of biological analytes in a biosensor has led to label-based techniques in which an additional molecule is attached to immobilized target molecules, viruses, or cells to enhance quantitative signal 41 . In a typical biosensor, the specific bioreceptors are immobilized on the chip sensing area to detect the targeted pathogens or proteins. Only the target biomolecules will be bounded to their corresponding biomolecular receptor upon introduction of analytes into the sensing area.
Notable examples of labels used in biosensing are dye molecules, a fluorescent tag, or enzyme.
Various types of bioreceptors-targets coupling mechanisms have been also demonstrated in Fig   3 including antibody-antigen binding, enzyme-substrate catalytic reaction and cDNA-DNA hybridization. These labels require sophisticated reagent selection and modification that in turn come with the drawback of perturbing the assay and making final detection a challenging task.
On top of that, labeling chemistry is both expensive and time-consuming. Thus, recent development in biosensing systems have been more intrigued by unlabeled or unmodified biomolecules (label-free biosensing) 35,41,46,47 in which native molecular properties like molecular weight and RI are utilized for sensing. Label-free detection does have its own shortcomings for example, it requires low non-specific binding 43 and sufficient signal to be generated upon targets binding 35 . However, its benefits such as providing real time analysis by simplifying assays and reducing time and number of steps required as well as eliminating experimental uncertainty far exceed its limitations 48 .
The sensing transduction signals in Optical label-free biosensing platform functions based on miniscule changes in refractive index resulting from the attachment of biomolecules to the immobilized bioreceptors. It is of vital importance to have a highly sensitive biorecognition layer on the transducer surface in a label-free optical biosensors 43,49,50 . It goes without saying that the biosensor final sensitivity and specificity is strongly dependent on the immobilized molecules and the accessibility of target analytes to them. Therefore, the optimization of sensing surfaces and their biofunctionalization strategies is a significant factor for an accurate label-free optical biosensor where the sensitivity and accuracy are highly necessitated 43 . The diverse range of target molecules and biosensor applications make it extremely difficult for a universal surface biofunctionalization procedure to be obtainable; hence, the procedure needs to be custom designed consequently. In a graphene-based Field Effect Biosensing(FEB), the surface is functionalized for protein immobilization with anti-Zika NS1  can perform a selective detection of the RdRp-COVID through DNA hybridization 29 . The surface functionalization of sensor surfaces has resulted in sensitivity improvement and suppressing the nonspecific bindings 29 . The virus-like particle absorption has resulted in more SPR peak shift on non-functionalized surfaces compared to functionalized one simply as there are a greater number of adsorption sites on it 53 .
Different structures have their own advantages and disadvantages. Elisa-based sensing requires additional secondary binding anti-body usage while it limits the application for fast detection requirements.

II. INTEGRATED OPTICAL BIOSENSORS
Here, we explore the most well-defined bio-photonic sensing mechanisms based on functionalized waveguides, interferometer, and resonance shift in microcavities.

A. Evanescent wave sensing: refractive index and absorption variation
Due to its immunity to electromagnetic interference (EMI), compactness and high selectivity, optical waveguides have attracted special attention as a basic biose nsing system 55,56 . The interaction of target molecules with bioreceptors on the surface leads to effective index and absorption coefficient change ( Fig. 4 (a)). The effective index change or loss change is a function of the concentration of biological or chemical targets on the surface.
Effective mode index change for a perturbed waveguide can be calculated through the variation method 47 : where ( , ) is the electric field, 0 is the free space impedance and represents the light wave power, and are the refractive indexes of the aqueous solution without analyte, and molecular adsorption layer, respectively.
The vertical distance from the interface z does reduce the strength of the evanescent electric field exponentially. The penetration depth (d) can be calculated from the incident wavelength (λ) and incident angle ( ) using the following formula: The surface sensitivity to specific target molecules is not suitable for general comparisons between sensors operating with different biosensing assays. On the other hand, by simply depositing a gold layer, the device concept can be used for a surface plasmon resonance, which can improve the LOD even further 71   In order to maximize the sensitivity of the waveguide-based biosensor, the speed of light can be reduced even further. For this reason, the group velocity of the input pulse can be reduced by designing the photonic crystal structure 35 . Photonic crystal waveguides slow down the speed of light and introduce other enhancement factors, thereby enhancing absorption-based sensing on the surface of the waveguide.

B. Integrated interferometer sensing
Integrated interferometer photonic is one of the most practical architectures for sensing applications. It is based on splitting the input beam into two arms through a Y-junction, one arm is completely retained as the reference arm, and the other arm contains the target. The interaction of electromagnetic waves on the sensing arm will cause a phase difference with respect to the reference arm, and recombination of the two beams in the output will cause constructive or destructive interference, see Fig 5 (a).
Output intensity of the MZI is described as follows 34,74,75 : where and are light intensity in reference and sensing arms, 0 is initial phase difference between two arms without external perturbation. The sensitivity of the MZI-based sensor is related to the phase sensitivity relative to the length of the sensor arm: In an imbalanced MZI, considering the phase matching condition, vis-a-vis the wavelength sensitivity, we can approximate the phase sensitivity of the MZI-based sensor: where, ∆ is the free spectral range (FSR) and is the spectral sensitivity.
Chemical and biosensing via Mach-Zehnder interferometers (MZI) have been widely exploited 76  The overall sensitivity of the manufactured interferometer sensor with a sensing length of 5 mm is reported to be 316π rad/RIU, and the extinction ratio can reach 18 dB. This method is promising for future commercial development. However, reliability of polymer needs to be further studied. Figure 5g shows the schematic structure of a spatially resolved resonant waveguide grating (RWG) for single cell detection. The sensor consists of (1) glass substrate, (2) a grating part, and (3) a waveguide with high refractive index. Owing to total internal reflection after the light guides in the waveguide, the light propagates out of the waveguide.
Note that the wavelength of the resonance is unique to the coupled light and is sensitive to the local reflection index, which is proportional to the density of the target analyte in the penetration depth of the biosensor. The mechanism of RWG device can be also considered in the resonance-based biosensors. Figure where dA is the thickness of the adsorbed layer, nA is the refractive index of adsorbed molecules, nC is the refractive index of cover solution and dn/dc is the change in the refractive index of molecules, which is proportional to the shift dλ in position of the resonance peak. The size of the resonance wavelength shift is proportional to the number of adsorbed biomolecules, thus providing a label-free method to quantitatively determine the target analyte.

Ring Resonators
Although high quality (Q) ring resonators can be achieved with a larger radius, the tradeoff between the Q and the free spectral range (FSR) limits the radius for a given FSR, which should be large enough for effective recognition of the sensing signal from the adjacent interference signals or for large scale on-chip multiplexing sensing applications. Wang et al. 82,83 proved through experiments that the Q-enhanced SWGMR was specially designed using a trapezoidal silicon column (T-SWGMR) ( Fig 6). According to the report, the SWGMR has a Q of ~5600 even with a large radius of 15 μm, smaller radius provide much higher Q.
Contrasted with conventional rectangular silicon pillars comprised of SWGMRs (R-SWGMRs), an asymmetric effective refractive index distribution is created, which can significantly reduce bending loss and thus increase the Q of SWGMRs.
The experimental results show that the applicable Q value of T-SWGMR with a radius of 5 μm is as high as 11,500, which is 4.6 times the Q value (about 2800) provided by R-SWGMR with the same radius, indicating that the propagation loss is reduced by 81.4%. To go one step further, Yan et al. 84 proposed a T-SWGMR biosensor and demonstrated the unique stable surface sensing characteristics through a demonstration of miRNA detection at a concentration of 1 nm (Fig 6b) In addition to utilizing the unique stable sensing characteristics of SWGMR and the enhanced Q of T-SWGMR, Chang et al. 69 showed a pedestal T-SWGMR biosensor that maximizes the mode volume overlap by implementing an asymmetric refractive index distribution along the vertical direction on the silicon-on-insulator (SOI) platform, thereby further improving sensitivity ( Fig 6.d). Both theoretic analysis and experimental proofs show that the volume sensitivity and surface sensitivity have been significantly increased by 28.8% and 1000 times, respectively. For streptavidin, a spectrometer with a resolution of 0.01 nm is used, and its LOD is about 400 fM. Owing to imperfect manufacturing process, experimental Q estimate of T-SWGMR with a radius of 10 μm and FSR of ~13 nm is 1800. The optimized SWGMR with symmetric coupling demonstrated by Huang; et al. 85 estimated Q to be 9800.

Microtoroid
Microtoroids are resonators with a Q of >10 8 and a small mode volume which can be fabricated on silicon using standard microelectronics techniques 86 . However, microtoroids need to be strictly aligned with the tapered fiber waveguide to achieve high coupling and cannot meet our needs for high-throughput multiplexing sensing. Vahala; et al. 86 demonstrates the possibility of detecting unlabeled single molecules and higher concentrations on a single platform (Fig 7a, b). According to reports, the quality of performing planar lithography is about 1.83×10 8 . The author reports that by using the IL-2 solution, the micro-ring sensor can provide a dose response of 10 -19 M to 10 -6 M and a working range of 5 aM to 1 µM. In another report for a microtoroid with a diameter of 90 µm, authors reported 87 that a measurement lifetime of 43 ns corresponds to an inherent quality factor of 1.25×10 8 (Fig 7c, d).  chip manufacturing process and sensor functionalization process will be the same as before.  Our photonic crystal microcavity not only has high sensitivity and low detection limit, but also can achieve dense integration of sensors due to its small geometric size. In Table 1 below, we compare our proposed PC microarray platform with commercially available benchtop systems for water monitoring such as inductively coupled plasma mass spectrometry (ICP-MS) and inductively coupled plasma optical emission spectrometry (ICP-OES, for metals), gas chromatography-mass spectrometry (GC-and GC-MS for organics) and ELISA (for all analytes with bio-signatures). We compared the technical advantages of our proposed platform with other platforms and showed that our platform can provide comparable sensitivity to existing desktop systems, while also being portable.

III. PLASMONIC OPTICAL BIOSENSORS
A plasmon can be described as a collective oscillation of a free electron or a quantum of plasma

A. Surface plasmonic resonance (SPR) sensor
Fundamentally, when the phase matching condition between the incident light and the SP wave guided along the metal/dielectric interface is reached, the incident light can be coupled to the surface guided mode. Note that the resonance condition between the incident light and the conductive electrons at the metal/dielectric interface with a fixed angle of incidence is only achieved at a specific wavelength. The guided light will be absorbed by the conducting electrons that resonates, which will significantly reduce the reflected light at that particular wavelength. Therefore, once the target molecule is attached to the functionalized metal film, the refractive index does change, causing a shift in the resonance wavelength. Consequently, SPR angle alteration can be characterized as the main sensing mechanism. Several coupling methods have been proposed, including a grating coupler, a waveguide coupler, and a prism coupler, but the prism coupling method has been used as a standard configuration based on the Kreichman configuration 114 . Figure 11. shows the schematic of the conventional SPR sensor configuration.
where is the angular frequency of the wave, c is the speed of light in vacuum, and are the refractive indices of the metal and dielectric. As aforementioned, the resonance condition is met when = , so we can calculate the SPR angle in the following equation: The sensitivity of the SPR devices are determined by the resonance shift with respect to the change of the refractive in the absence and presence of the target analyte [118][119][120][121][122] : where Δ is the resonance wavelength shift and Δ is the change of bulk refractive index including the target analyte.

B. Localized SPR sensor
On the other hand, nanostructures in conductive thin films are among the essential building blocks of LSPR plasmonic biosensors (see Fig.12). These nanoscale geometric/periodic lattice factors bring huge advantages over conventional SPR devices.
Contrasted with SPR occurring along the propagation surface, the attenuation length of the local electromagnetic field is much shorter. These strict restrictions, with a shorter subwavelength structure, can achieve ultra-low mode volume resonance, making it sensitive to environmental refractive index changes, which are particularly helpful for the detection of tiny biological molecules. Also, an incident light can be directly coupled to SP wave on the conductive structures without any external couplers, e.g., prism or gratings, which ameliorates the complexity of the entire system and enables the sensor miniaturization 123   Moreover, LSPRs can be utilized for various types of resonance modes and detection methods, including surface-enhanced Raman spectroscopy (SER) 28,112 , photoluminescence/fluorescence 50,111 , and mid-infrared spectroscopy 106 , by tuning the resonance wavelength for a specific light-matter interaction. Table 2 shows the comprehensive comparison between the conventional SPR and LSRP biosensors.

Resonance shift sensing
where and FWHM are the wavelength and full-width half maximum of the resonance peak, respectively. To enhance the sensing performance, a higher Q value is desirable because of the reason that sharper peaks with high Q values are much easier to detect. Considering all these factors, the inherent detection limit (ILOD) of the resonance displacement sensing device can be defined as follows 80,129 : which indicates that both the higher sensitivity (S) and Q factor are required to minimize the limit of detection of the sensors.
Although these so-called hot spots provide higher sensitivity for LSPR biosensors, their performance is greatly limited due to the basic limiting factors of ohmic losses in metal surfaces. In other words, compared to other photonic biosensors, the absorption loss in the conductive nanocavity leads to a low Q value, so research has been conducted to achieve lowloss devices by using advanced materials or optimizing the geometry of metamaterials. On the other hand, the concept of a plasma perfect absorber (PPA) sensor was introduced to overcome this intrinsic limiting factor [130][131][132][133][134] . Figure 13 shows the typical configuration of PPA sensor consists of periodically arranged metallic nano antennas (metamaterial) on top and thin metallic 'mirror' layer on the bottom separated by dielectric spacer 132 . The basic concept is to have a perfect absorbance at the operating wavelength and make a 'zero' transmittance by maximizing the metamaterial losses; in other words, the losses are served as an advantage in the PPA sensors. In this structure, most of the incident light at the operating wavelength is absorbed by top nano antennas operating as a resonator through impedance matching, and the metallic bottom layer act as a 'mirror' to eliminate the transmittance. As a result, the reflectance of light can be characterized for sensing as in figure   13, and the figure of merit (FOM ) is defined as below 133 :

Plasmonic perfect absorber
where ( )/ ( ) is the relative intensity change of reflected light at a fixed resonance wavelength , which is induced by a refractive index change ( ).
Moreover, it has been shown that the perfect absorption (>99%) of incident light at working wavelength can be remained over a wide incident angle and insensitive to the polarization (TE/TM) of incident light 133 . Amongst many types of optical based molecular absorption spectroscopy platforms, SEIRA spectroscopy for LSPR devices has been shown its great promise for detecting thin layer of surface-bound nano-molecules due to its tight confinement of surface plasmons on metallic nanostructures, which can significantly enhance the IR absorption of small molecules.

SARS-CoV-2 sensing Application
Here, we review the most up-to-date advances, especially for the coronavirus sensors in plasmonic domain, and introduce well-established plasmonic SARS-CoV-2 biosensing systems. Researchers have demonstrated that using SPR/LSPR-based sensors and corresponding binding biological receptors can effectively and selectively detect coronavirus 103,105,113 (Table 3). Moreover, several researches have already reported SARS-CoV-2 sensing results as in figure 16 29,126 .  Furthermore, several researches reported that the sensitivity and the signal-to-noise ratio (SNR) of conventional enzyme-linked immunosorbent assay (ELISA) or fluorescencelinked immunosorbent assays (FLISA) tests can be significantly improved by applying the 'add-on' plasmonic particles without altering their workflow 50,103 . As the ELISA test is widely used for precise SARS-CoV-2 detection 103 , plasmon enhanced ELISA/FLISA tests can be applied to COVID-19 sensing as well.
Although aforementioned SARS-CoV-2 sensing applications 29,126 have shown great performances of sensitive and selective sensing, a huge potential for more sensitive, accurate and fast on-chip sensing with less complex system is still remained in LSPR biosensor domain.
For example, the sensing systems in Fig. 16 29,126 require the prism coupler to couple the incident light into SPR device with an accurate incident angle. It requires very sensitive alignment of optical devices which makes the overall system complex and hard to be integrated with sources and detectors. However, as described earlier in Table. 2, the incident light can be coupled into LSPR sensors directly without the external couplers, and this normal-incident angle can make the alignment easier, in turn mitigate the complexity of the system and make the possibility of fully integrated on-chip sensing; moreover, due to the capability of sensor miniaturization through LSPR nanostructures, label-free, real-time, and parallel detection with multiple channels with high-specificity are achievable. Furthermore, improving the sensitivity by applying advanced materials has incited a great interest for various optical biosensor applications. For plasmonic biosensors, the ultrasensitive graphene and 2D material enhanced SPR devices have been reported as shown in Fig. 16 (h) and (i) 142,143 , and the experimental sensing result with LOD value approaching 1 atto M has been shown 142,144 . Accordingly, the advanced material enhanced LSPR biosensors are anticipated to enable the possibility of highly sensitive, accurate and fast point of care lab-on-a-chip integrated sensor with unprecedented high sensitivity. The detailed discussion of emerging nanomaterials for optical biosensors are described in section V.

IV. SARS-COV-2 BIOSENSOR: DESIGN AND IMPLEMENTATION
To develop an accurate estimate of COVID-19 biosensing functioning mechanism, a simulation model need to be first designed. Here, in a proposed simulation model, COVID-19 is approximated to be a solid sphere core containing RNA covered with a membrane protein with radiuses of r1 and r2, respectively (Fig. 17a) 45 . Thus, the effective RI of the virus is calculated by taking a volume weighted sum of the two refractive indices: where n1 (V1) and n2 (V2) are the total RI of the RNA and the membrane protein volume, respectively. As the RI of the virus is determined mainly by material composition rather than its geometrical size, η is a constant value for the same kind of virions. (ηCOVID-19 =1.25 average value of several measurements of transmission electron microscopy (TEM) pictures 145 ). We use the SWGR design to simultaneously take advantage of the enhanced binding surface and strong light-substance interaction. As shown in Fig. 17 c and e, the energy mode is distributed between the gratings as well. In order to further improve the SWG waveguide functioning in the sub-wavelength range, the grating period Λ, the waveguide width w, and the fill factor are designed to be 230nm, 1.23μm, and 0.5μm, respectively. For the SWGR, the radius R is set as 5μm with the corresponding FSR 57 of 25nm at 1550nm. Here simulation system includes a 220nm-Silicon top-layer with a 3μm buried oxide (BOX) wafer and a liquid solution, with the refractive index of nclad to be 1.35 31 . Adopting our previous designs features 38,39 , we optimized a high-Q SWGRs by utilizing a trapezoidal (T) silicon pillars and reducing bending loss by ~50% compared to a conventional rectangular silicon pillar. We therefore set the SW waveguide width to be 0.5µm (correlated to the fundamental mode of transverse electric (TE)) and studied the effect of the trapezoidal width. It is noted that to obtain the lowest bending loss of the T-SWG waveguide, we employ the particle swarm method for the optimization process.
Three parameters (w, A1, A2) are optimized and are defined as the width, the tuning factor of the outer and inner filling factor of the SWG, respectively (as shown in the inset figure of Fig.   17b). Considering the limitations of the design for fabrication, the slot between gratings are pre-set to be larger than 60nm, thus A1 and A2 are limited to be (1, 2) and (0, (1-60nm/Λ)/f), respectively. At the same time, to keep the SWG working in subwavelength regime (Λ<< /2neff), Λ is safely set to be 230nm and f is simply set to be 0.5 with no optimization.
Furthermore, to make the SWG waveguide work as a single or few mode waveguide, the width of the gratings is set less than 2μm. All in all, the ranges of w, A1, and A2 are set to be (0.5μm, 2μm), (1,2), and (0, 0.522), respectively. The FOM is defined to achieve the lowest bending loss with the bend radius of 5μm. Based on the optimization measures taken, we finally achieved a bending loss as low as 0.0279 dB/cm with the optimized (w, A1, A2) = (1.23µm, 1, 0.522). By adjusting the coupling gap between the insertion SWG waveguide and the designed SWGR, the Q can be as high as ~50000 (the resonance at 1557.6nm) with a broad FSR of 25 nm, as shown in Fig. 17a. We also optimized the 10µm radius SWGR (not shown in the Fig.17f, achieving a loaded Q of ~75000 (the resonance at 1552.1nm) with the FSR of 11nm with (w, A1, A2) = (1.23µm, 1, 0.522), at the same waveguide-ring cross-coupling coefficients. Needless to say as quality factor of the ring becomes higher, the fabrication tends to be more challenging.
Bulk RI sensitivity (shown in the Fig.11f inset figure) in the buffer solution is calculated to be = = 400 / . Thus the iDL can be calculated as low as ~7.5e -5 RIU. Note that iDL performances can be further improved by exploiting a larger radius ring or by further achieving the critical coupling condition given the predictable higher Q, while making tradeoff between the performance and the sensor size or the resonance peak extinction ratio.
Surface sensing: To evaluate the specific sensing ability of the proposed device for COVID-19, surface sensing performances are analysed by considering the device immersed in buffer solution, bonded by several surface layers (generated in the sensing preparation process) including the ~2-3nm surface oxide layer, ~10nm functionalization layer and bonded antibody (protein layers), and the bonded virus particles layer in detection process. In simulations, the preparation process generated layers are further simplified to be a uniform layer (RI: 1.45) with a thickness of 15nm, and the bonded virus layer is simplified as a uniform layer with a thickness of 125nm (the maximum diameter of the COVID-19 virus) (Fig. 17a). It is noted that the equivalent RI of the virus layer (nbinding) depends on the number of bonded virus, which is a function of the virus concentration and binding processing time, and is dominated by the concentration in real sensing process with a given binding time. Thus, the SWGR sensing performance can be evaluated by calculating the nbinding response of the device, with the nbinding ranging from 1.35 (no binding) to 1.5 (full binding). Simulation results in Fig.17f shows the functionalization and the full binding process induces a shift of 3.41nm and 1.14, respectively.
The obvious simultaneous measurable shifts in the FSR range ( < ) and experimental values ( ≫ 1 ) indicates the promising potentials of the proposed device in detecting the COVID-19 virus or simply being as a chemical/bio-sensor in future practical applications. with the bulk RI sensing shows a sensitivity of 400nm/RIU, and surface sensing shows the total wavelength shifts of 3.41nm nm and 1.14 nm after the functionalization process and the full binding process, respectively.

V. TWO-DIMENSIONAL MATERIALS OPTICAL BIOSENSORS
Latest major advancements in preparation, development, and utilization of new lowdimensional materials has been attractive for development of modern miniaturized biosensors and immunosensors. Graphene and its analogous 2D materials such as transition metal dichalcogenides (TMDCs), carbides and nitrides(Mxenes), hexagonal boron nitride (h-BN), black phosphorus (BP) and transition metal oxides (TMOs) have attracted great attention to be used as transducer due to combined high sensitivity and selectivity for biosensors.

A. Graphene and graphene oxide (Gr and GO)
Graphene has been regarded a revolutionary material ever-since its first introduction in 2004 146 , given its extraordinary optical and electronic properties 32,[147][148][149][150][151][152][153][154][155][156][157][158][159][160][161] . Since then, graphene has also shown an immense potential in different applications and a great deal of graphene based biomolecular sensors have been specifically developed by paying especial attention to its biocompatibility and high specific surface area 45,[162][163][164] . On top of that, unique and ideal optical properties such as broadband and tunable absorption and polarization-dependent nonlinear optical effects make graphene a promising candidate to be employed for optical based biosensors (Fig 18). The introduction of advanced biosensors through graphene electrical and optical qualities in general has delivered extraordinary sensitivity, detection level, resolution and response time in many devices (Fig 18. c and d) 32,106,151,152,154,[163][164][165][166][167] .
Point-of-Care biosensors has also shown enhanced sensitivity in graphene based electrochemical biosensors as well 168 . While pristine graphene has seen applications in biosensors devices, its derivative GO has been subjected to a wealth of investigation for rapid detection, disinfection of pathogens and enzyme assays, making it a key material for a variety of biomedical applications. Graphene oxide has been a suitable precursor for graphene and its biosensors applications especially due to its attractive distinctive properties like good water dispersibility, facile surface modification and to be more specific photo- fM with the potential to show even higher sensitivity values 169 . They showed the high electron conduction and higher surface area in spherical morphology has been specially effective to improve the sensitivity and limit of detection 169 . Graphene unique electrical properties has also been exploited effectively to develop different transistor based label-free biosensors including COVID-19 detection system 39 (fig 18.a). Aside from the field-effect-transistor-based graphene biosensor, which relies mainly on current changes, providing easier mass-scale production with satisfying sensitivity, it's limited sensing capability along with being damaging to living cells make its application limited compared to analogous optical ones 170 [ Table 3].

B. Graphene Surface Plasmon Resonance
Researchers have investigated plentiful ways of enhancing the sensitivity of the SPR sensor, including but not limited to the use of resonant structures such as metal nanoparticles and optical gratings. As shown before, in a typical SPR or LSPR biosensor, a thin metallic film like Ag or Au gets deposited on prism to separate it from the sensing area. The deposited metallic film brings out the propagation of surface plasmon at visible light frequency. Gold is preferred as it provides better resistance to oxidation and corrosion in different environments. The intrinsic defects in gold and silver-based biosensors like oxidization of metal, poor adsorption to biomolecules and hence limited sensitivity and accuracy leads researchers to seek methods of alternatives. In view of the defects of biosensors based on gold and silver film, graphene-based biosensor has been developed. Graphene provides a highly sensitive non-oxidizing receptor substrate to analytes 152,166,175 . Moreover, graphene also helps to adsorb biomolecules better, because of π-π stacking, which increases the system's affinity for these molecules (Fig 19)  Schematic of a SPR biosensor functioning mechanism(Reprinted 44 ). Photocurrent is too small due to low absorption rate of graphene The graphene plasmonic nano-islands can demonstrate nonlinearities two orders of magnitude higher than the their non-graphene counterparts of equal size 174

VI. CONCLUSION
Emerging pandemics and epidemic diseases like COVID-19 brings out a high demand in advancement and research in medical detection and treatment methods. The optical biosensors provide a fast detection (< 1min) of such a virus at very low concentrations (~1 fM). However, they need to be designed and functionalized to be the most absorptive to the target analyte. The ideal label-free biosensor is cheap, disposable, or reusable, compact, and semi-automatic. Although most efforts in biosensors have been focused on protein biomarkers, other targets such as small molecules and nucleic acids are crucial in expanding the application of biosensors including optical ones. A common challenge for optical biosensors is to reach the capability of performing the measurement in real complex samples, avoiding, or limiting the sample preparation phase. Developing label -free biosensor is aligned with that purpose as the need for on-site detection techniques is boosting as the world post-COVID-19 pandemic will never be like before.

Acknowledgment
This research team is supported by AFOSR MURI silicon nanomembrane research center

Data Availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.