Recent advances in environmental and clinical analysis using microring resonator-based sensors

Abstract

Progress in the development of biosensors has dramatically improved analytical techniques. Biosensors have advantages over more conventional analytical techniques arising from attributes such as straightforward analyses, higher throughput, miniaturization, smaller sample input, and lower cost. Microring optical resonators have emerged in the area of optical sensors as an exceptional choice due to their sensitivity, ease of fabrication, multiplexity capability and label-free detection. In this paper, the sensing principle of these sensors is described. In addition, we summarize and highlight their most recent and relevant applications in environmental and clinical detection analysis.

Graphical Abstract

Introduction

More than half a century since the emergence of the glucose biosensor [1], biosensors have revolutionized the field of chemical and biological analysis. Offering excellent sensitivity and selectivity, these devices can provide valuable information about the concentrations of targets within easy-to-use formats [2]. Biosensors are analytical devices that convert a biological interaction or recognition event into a measurable electronic signal. They generally consist of three parts (Figure 1A): a recognition element, a transducer, and an electronic readout. The recognition element is usually a biomolecule such as an enzyme, antibody, or nucleic acid, although other target-selective capture agents may be used. Biosensors are classified by the diverse types of transduction modalities, which commonly include optical, electrical/electrochemical, or acoustic methods.

Figure 1.

A. Schematic representation of the elements constituting a microring biosensor device: analytes, biorecognition elements, transducer and signal readout. B. Light only couples into the microring under conditions of optical resonance. In the all-pass configuration, resonances are wavelengths at which the intensity of light transmitted past the resonator is heavily attenuated due to localization on the resonator. C. The evanescent wave extending from the waveguide boundary samples the local refractive index. In this way, changes in local index—often due to analyte binding to microrings modified with specific capture agents—leads to a shift in resonance wavelength that can be measured in real-time to facilitate analyte detection. D. In the add-drop configuration, resonances are measured as an increase in light intensity in the output waveguide as it is transmitted via the microring resonator.

Optical sensors constitute one of the most conventional types of biosensors because they can provide direct, real time measurement of molecules, occasionally in label-free formats [3]. Optical sensors function through the interaction of an optical field with a recognition region to reveal binding-induced changes in optical properties. Label-free optical sensors can detect analytes without the need for secondary target labels (e.g. fluorophores or enzyme tags). Common label-free optical sensing technologies include surface plasmon resonance (SPR), interferometers, and whispering gallery resonators, among others. These sensors similarly are able to confine light at an interface with the surrounding (sample-containing) medium. At this boundary an evanescent field extends into the surrounding region such that the local optical properties are sensitive to binding interactions at or near the surface [4].

Given the extensive literature in the theoretical characterization of these devices, we point the reader to more exhaustive reviews in this aspect [5,6]. Briefly here, whispering gallery mode (WGM) biosensors confine photons in a path circumscribing the circular cavity. These photons can recirculate many times around the cavity allowing for small changes in the recirculating path to be detected with high sensitivity. Facilitating these analyses are a tightly held resonance condition that is established based upon a constructive interference condition at the junction between the coupling waveguide and circular path of the cavity (Figure 1B). This resonance condition is defined mathematically by the following formula:

$$m\lambda =2\pi r{n}_{eff}$$

where m is a non-zero integer, λ is the wavelength traveling in the waveguide, r is the radius of the cavity and n_eff is the effective refractive index sampled by the optical mode traversing the cavity. Thus, changes in the local refractive index (a component of n_eff) will lead to a shift in the resonance wavelengths supported by the cavity (Figure 1C). Most WGM sensors operate by measuring the shift in wavelength due to binding-induced changes in the local refractive index though sensors based upon measurement of the broadening [7] and splitting [8] of resonances have also been successfully demonstrated. The sensitivity, S, of these sensors can therefore be defined as:

$$S=\Delta \lambda ∕\Delta n_{eff}$$

WGM sensors have shown to have bulk sensitivities up to several hundred nm per refractive index units (nm/RIU) [9]. These measurements are comparable to other optical sensors based on refractive index measurements. However, more practical when talking about sensors it is the use of responsive layers and therefore the sensitivity of the system does not only depend on the bulk refractive index change but more in the recognition and noise levels of the sensor [5,10].

One important parameter used to characterize these sensors is the quality factor, Q. This factor is a measure of the sharpness of the resonant dip relative to its central frequency. Remarkably to the sensing properties, this parameter is related to the number of recirculation of the photon in the resonator and therefore to the confinement of the light. Higher Q factors are related to sharper resonant dips and higher sensitivities. Q factors between 10³ to 10¹⁰ have been reported [11,12]. Improvements in the Q-factors have been achieved by altering the design configuration such as the use of photonic crystal structures inside the microcavities [13] and the use of materials and fabrication techniques that reduce the roughness of the surface [14].

WGM sensors have been produced in several different geometries, including micro- spheres, toroids, disks, and rings[15]. Microspheres and toroids offer a great confinement of the light, and possess greater Q factors. However, they required complexed optical and mechanical systems for the alignment of the light source and readout equipment and their three dimension structure is harder to fabricate than other planar geometries such as disk and ring resonators [16]. Microrings have been the most common geometry for biosensing. The ring geometry, usually attached to planar underlying substrates, provides many advantages in the ease of construction using CMOS- (Complementary metal–oxide–semiconductor) compatible semiconductor fabrication approaches. The planar substrate geometry is also amenable to the creation of multiplexed sensor arrays. Although planar substrate microring arrays have been most commonly reported, optofluidic microring resonators have been impressively developed. Optofluidic microring resonator devices often consist of a thin walled capillary, wherein the cross section acts as a whispering gallery mode sensor. Analytes flow through the inside of the capillary interacting with the evanescent field from the inner surface. The ease and relatively low cost of integrating the microfluidics with this sensing modality is particularly advantageous [17].

Ring resonators can be configured as all-pass filters or add-drop filters [18]. In the all-pass filter configuration, which only requires a single linear waveguide, shifts in optical properties are simply recorded as changes in the transmission spectra measured past the resonator (Figures 1B-C). In contrast, the add-drop filter configuration utilizes two linear waveguides positioned on either side of the resonator. Light is coupled into the resonator via the input waveguide and the signal measured (or “dropped”) into the output waveguide is monitored as a read out (Figure 1D). In addition to these two configurations, there have been recent advances such as cascade ring resonators [19] or coupled resonators optical waveguides [20] that employed more complex configurations to increase the sensitivity of the rings.

Microring resonators can be easily fabricated from semiconductor materials that have refractive indices such that the resulting waveguides feature good light confinement. Common materials systems include silicon (typically silicon-on-insulator), silicon carbide and silicon nitride [21]; however, other materials including polymers have been demonstrated as potential lower cost alternatives that offer unique chemical properties [22]. Among other innovative materials are n-doped semiconductors that can simultaneously support resonant optical modes and electrochemical interrogation at the sensing surface. In these electrochemical ring resonators, the electrochemical activity of this interface can also enable site-selective immobilization of biomolecules to create sensor arrays [23].

While this review focuses almost exclusively on progress from academic laboratories that are driving new applications and advances in this field, it is worth noting that microring resonators have already been commercialized by Genalyte, Inc. [24]. Genalyte has demonstrated the ability to detect autoantibodies that are diagnostic signatures for a range of connective tissue disorders[25,26], and their commercial products are finding early adoption in clinical laboratories. Given the focus on clinical lab-based testing, the full integration of on-chip optical sources and detectors has not been a commercial focus of Genalyte, and as such the instrument is benchtop and not yet amenable to portable detection applications.

Herein, we review the most recent applications of ring resonators for environmental and clinically-relevant sensing. Key figures of merit compared include type of analyte, detection strategy, limits of detection, and demonstrated ability of operation within complex sample matrices.

1.1. Environmental analysis using microring resonators

In environmental sensing applications, microring resonators have been utilized to detect diverse chemical species including gases, heavy metals, pesticides, explosives, biological toxins, and whole microorganisms. Selected examples are described in the following text, and also summarized in Table 1 and Figure 2A.

Table 1.

Recent demonstrations of microring resonators for environmentally-related analysis.

Analyte	Detection strategy	Matrix	LOD	Reference
DMMP (organophosphorus agent simulant)	A chemoselective coating that changes absorption spectra of MRR	Air	2 ppb	[27]
Hydrogen gas	Palladium disk inserted inside of microring cavity	Air	11.038 nm/% H2	[28]
Mercury and lead ions	Mesoporous silicate matrix	Water	1 ppm	[30]
DMMP	Selective recognition of the targeted DMMP molecule by specifically modified proteins immobilized on photonic structures	Air	6.8 ppb	[29]
4-methylumbelli-feryl phosphate (organophosphate simulant)	Selective partitioning of organophosphorus compounds into polymer brushes grown directly off of microrings	Buffer	0.1 mM	[31]
Aflatoxin M1	Aptamer and Fab’ recognition	Buffer	5 nM	[32]
Ricin	Single Domain antibodies	BSA-containing buffer	200 pM	[33]
Lectins from Aurelia Aurantia and Sambucus Nigra	Surface functionalization with glycan receptors	Buffer	7 pM and 86 pM	[34]
Bean pod mottle virus Virus	Antibodies against the virus	Leaf extract dilutions	10 ng/mL	[35]

Figure 2.

A. Applications of microring resonators in environmental monitoring with examples in: A.1. Gas detection: Schematic architecture of the hydrogen sensing device with inner Pd disk [28]. Reprinted from [28] Copyright 2017 Optical Society of America. A.2. Aqueous solution detection: Tetrasulfide-functionalized mesoporous silica film over SOI microring for heavy metal ions [30]. Reprinted from [30] Copyright 2015 Optical Society of America. A.3. Biological toxins: Microrings functionalized with single domain antibodies for the label free recognition of ricin in buffer samples [33]. Adapted with permission from [33] Copyright 2013 American Chemical Society. B. Applications of microring resonators in clinical analysis with examples in: B.1. Nucleic acid detection: Direct detection of miRNA without labeling. Quantification based on the slope of the binding curve [36]. Adapted with permission from [36] Copyright 2010 John Wiley and Sons. B.2: Protein detection: Direct multiplex detection of phosphoproteins using from cell and glioma tissue lysates [21]. Adapted with permission from [21] Copyright 2015 American Chemical Society. B.3. Whole microorganisms: Test and control microrings on a resonator chip, showing specific bacterial binding [37]. Adapted with permission from [37] Copyright 2007 Elsevier B.V.

For the detection of gases, a demonstrated strategy combined microrings with materials that are chemically responsive to the gas (Figure 2A.1). One example utilized microrings modified with a hyperbranched carbosilane polymer that changed the absorption spectrum of the microring upon exposure to trace levels of phosphonate ester nerve agents [27]. Another example introduced a Palladium microdisk within the microring architecture that expanded upon hydrogen exposure, leading to a resonance shift accompanying hydrogen incorporation into the structure. Using this strategy, the reported sensitivity was as high as 11.038 nanometer disk expansion/% hydrogen, resulting in a ~23% enhancement compared to other hydrogen WGM sensors [28]. Yet another example of environmentally-relevant target analysis involved the detection of dimethyl methylphosphonate (DMMP), a precursor of the nerve gas sarin that is often used as an analyte for testing. In this case, the microring surface was functionalized with bovine odorant-binding proteins (b-OBP) that had been genetically mutated to have a high affinity for DMMP. This approach yielded a limit of detection (LOD) of 6.8 ppb without the need for any sample pre-concentration [29].

Microring modification approaches have also been shown to be effective in detecting environmentally-relevant targets in aqueous environments (Figure 2A.2). Microrings have been modified with a mesoporous silicate to extract heavy metals from aqueous solutions. The uptake of mercury and lead into the mesoporous silica matrix led to changes in the refractive index of the microrings, resulting in a measureable change in the sensor signal down to 1 ppm in concentation [30]. Microring resonators have also been modified with polymer brushes grown directly from the sensor surface via atom transfer radical polymerization. Different polymer brushes can impart chemical selectivity via partitioning, as was demonstrated for several small molecule organic analytes, including an organophosphorus simulant. Furthermore, the magnitude of the resonance shift was found to be directly related to target concentation, providing an approach for quantiative monitoring of water contaminants [31].

Microrings have also served as a useful technology for detecting biological toxins in water and food samples (Figure 2A.3). One target has been aflatoxin M1, a mycotoxin from Aspergillus that can be found in milk products. Microrings were functionalized with aptamers or antigenbinding F_ab’ fragments to selectively detect the toxin in milk with limits of detection of 5 nM [32]. Another naturally occuring and potent toxin, the lectin ricin, which can be isolated from castor beans has also been detected using a microring-based approach. Sensors functionalized with single-domain antibodies were found to be selective for ricin over other similar toxin compounds with a limit of detection of 300 pM in just a 15 minute assay [33]. Silicon nitride microring resonators have also been employed for the multiplexed detection of of Aleuria Aurantia Lectin (AAL) and Sambucus Nigra Lectin (SNA) with glycan-functionalized sensors giving limits of detection of 7 pM and 86 pM, respectively [34].

Microrings have also been explored for the detection of whole organisms in environmental samples. In one example, Bean pod mottle virus, one of the most common viral soybean pathogens that can limit crop yields when fields are infected, was detected using microrings functionalized with an antibody that recognized an outer capsid glycoprotein on the virus. Using this strategy, a limit of detection of 10 ng/mL was achieved for whole, intact virus particles, and, importantly, this assay was shown to be able to detect the virus directly from ground soybean leaf samples in less than 45 min [35].

1.2. Microring resonators for analysis of clinical samples

Applications in multiplexed clinical diagnostics have fueled much of the development of microring resonators as a biosensing platform, and not surprisingly they have been broadly applied to many classes of clinically-relevant biomolecular targets. By attaching different types of target-specific capture agents to localize antigens to the sensor surface microrings have been demonstrated for the quantitative detection of different classes of targets—including nucleic acids, proteins, smaller biomolecules, and whole microorganisms—in the context of human health applications. Selected examples are described in the following text, and also summarized in Table 2 and Figure 2.B.

Table 2.

Recent demonstrations of microring resonators for clinically-relevant analysis.

Analyte	Detection strategy	Matrix	LOD	Reference
Insertion sequences: IS6110 and IS1081 of Mycobacteriu m tuberculosis H37Rv	Recombinase Polymerase Isothermal amplification and ring complementary hybridization	Extracted DNA from sputum samples	1 fg and 10 fg of genomic DNA	[40]
9-plex miRNA panel	Asymmetric PCR and ring complementary hybridization	Extracted RNA from cells and tissue biopsies	2 pM	[41]
12-plex (Phospho)proteins panel	Antibody capture probe and signal amplification with an enzymatic turnover of precipitate on the surface	Brain tissue and cell lysates	0.6 pM IgG. Analysis focused on the relative expression of phosphoproteins upon different treatments	[21]
7-plex cytokine panel	Antibody capture probe and signal amplification with an enzymatic turnover of precipitate on the surface	Supernatant from activated periph eral blood mononuclear cells	0.5-65 pg/ml	[49]
Viral glycoproteins (Ebola, Marburg virus, dengue)	Antibody capture probe and signal amplification with streptavidin beads	Blood and Saliva	1.6-39 ng/ml	[50]
Thrombin	An aptamer that binds thrombin without further amplification	Buffer complemented with BSA	50 ng/mL	[51]
Testosterone	A layer of molecularly imprinted polymers for the recognition of testosterone	Water	48.7 pg/mL	[52]
Glucose	Capillary-based microring resonator	Aqueous buffer	0.035 mM	[53]
Escherichia coli	Microrings functionalized with monoclonal antibodies against the bacteria	Buffer	105CFU/ml	[37]
CD4+ and CD8+ lymphocytes	Antibodies immobilized in the inner surface of the capillary microring sensor	Buffer	200 cells/ μL	[54]

The detection of nucleic acids commonly relies on the functionalization of the sensor surface with oligonucleotide probes that are complementary to the sequence of interest (Figure 2B.1). Although direct detection of nucleic acids has been reported [13,36,38,39], recent improvements in the LOD and required sample input have led to the integration of DNA replication by Polymerase Chain Reaction (PCR) or isothermal amplification and other PCR-free signal amplification schemes. For example, microring resonators were combined with recombinase polymerase amplification to detect the insertion DNA sequences from Mycobacterium tuberculosis in sputum samples [40]. This amplification strategy allowed for to the multiplex detection of the two tuberculosis biomarker sequences down to 3.2 or 12 copy numbers per the 10 μL reaction volume. Similarly, the combination of DNA amplification and microring array-based detection demonstrated successful multiplexed miRNA detection [41]. In this work, miRNAs sequences were amplified via asymmetric PCR, a variant of PCR that produces single-stranded DNA products. By functionalizing arrays of microrings with complementary hybridization capture probes, multiplex expression of a nine miRNA panel from glioma patients were compared to a healthy control using a total RNA sample input of 10 ng. Importantly, the combination of asymmetric PCR and microring resonators have also been utilized in the analysis of long-non coding RNAs, lncRNAs, showing a good agreement with conventional qPCR[42]. On the other hand, low limits of detection have also been possible without the need of gene replication by adding other signal amplification strategies. For example, nM levels were reported for the multiplex detection of miRNA by adding complementary biotinylated primers and horseradish peroxidase amplification signal[43].

Protein-detecting immunoassays have been realized on microring resonators often using antibodies for the recognition of the target epitopes (Figure 2B.2). Although microrings possess exquisite sensitivity, this class of assays have also benefited from the use of secondary amplification (similar to Enzyme-Linked Immunosorbed Assays (ELISA)) for enhanced selectivity and improved LOD. Early reports using antibody functionalized microrings demonstrated the label-free detection of carcinoembryonic cancer (CEA), a glycoprotein secreted in blood and has been established as a biomarker for many human cancers [44], as well as the creation of relatively simple label-free, multiplexed detecting arrays[45].

To increase the selectivity of protein biomarker assays—particular those targeting complex human sample matrices—tracer antibodies have been employed in sandwich immunoassay formats[46]. Beyond the added selectivity, these assays also present strategies for improving attributes such as assay dynamic range and limit of detection[44]. Candidate signal enhancement strategies include layer-by-layer biorecognition, which was applied to the analysis of cancer biomarkers in serum[47], and an enzymatic enhancement method that generated a spatially-localized precipitate on microrings that yielded subpicogram per milliliter limits of detection for inflammatory protein targets [48]. This enzymatic enhancement strategy has been demonstrated as robust for multiplexed protein detection from within clinical matrices, including the analysis of phosphoprotein levels from brain cancer samples [21] and identification of cytokine signatures in secreted cell supernatants that have promise for the diagnosis of latent tuberculosis infection [49]. Another signal enhancement strategy applied to microring resonator-based detection of viral glycoproteins in saliva and blood involved streptavidin-coated beads [50]. And in addition to antibody capture agents, microrings presenting immobilized aptamers have also been utilized for detection, including the detection of thrombin and IgE with limits of detection of 50 ng/ml [51].

Microring resonators have also been demonstrated for the detection of clinically relevant small molecules such as hormones and glucose. A layer of molecularly imprinted polymer (MIP) was used to analyze for testosterone in aqueous samples down to 48.7 pg/mL [52]. The MIP layer is interesting for analytical applications due to their potential for improved stability compared biological-based capture agents, such as antibodies. Optofluidic ring resonators have also be utilized to quantitate glucose with a LOD of 35 μM, which is one order of magnitude lower than nominal clinical ranges [53].

Microring resonators have also been applied to the detection of whole pathogenic organisms and mamallian cells of clinical relevance (Figure 2B.3). An early example included the detection of E. coli using microrings functionalized with monoclonal antibodies against the bacteria, demonstrating a LOD of 10⁵ CFU/ml [37]. CD4+ and CD8+ lymphocytes were also detected at 200 cells/μL concentration with an optofluidic microring resonator that was, again, facilitated by immobilization of target-specifc antibodies on the resonator surface [54].

Conclusions

Over the past 10-15 years, microring optical resonators have emerged as one of the most promising and multiplexable platform for clinical and environmental analysis. Attributes such as amenability to mass production (for silicon photonic resonators, in particular) and microfluidic integration (for optofluidic resonators) give these technologies inherent advantages for analyses that need to be performed in high numbers and from small sample volumes. These sensors can act in label-free assay formats; however, for many applications requiring high selectivity and low limits of detection, more complex assays requiring labels have been successfully employed. The myriad of analytical targets that have been demonstrated to be analyzed using microrings is vast—largely on account of the fact that any target that can be physically localized to the surface via a capture agent that provides selectivity is theoretically detectable. This review has tried to balance a historical context for technology development with a focus on recent advances that demonstrate new capabilities as well as important applications of real-world relevance. The microring transduction technologies themselves have become quite mature; however, there are remaining challenges for the field that increasingly include improved methods of sample delivery and integration with complementary technologies in such a way to facilitate hyphenated analyses that also include, for example sample pre-treatment via chromatographic methods[55].