Low-cost biosensing systems based on integrated photonic devices

The development of lab-on-a-chip (LoC) devices able to perform complex sample analysis quickly and easily, and at reduced cost, is crucial for many different fields of application. Examples include medical diagnostics, food and water safety control, and drug discovery. Transduction elements for these LoC devices based on integrated photonic technology are particularly promising. Their extremely high sensitivity and reduced size make them well suited for detecting target analytes (molecules of interest) at extremely low concentrations without the use of any type of marker (i.e., label-free). Hundreds or even thousands of these sensing elements can be arrayed on chips smaller than 1mm², enabling both multiparameter analysis and more complete and valuable information.

CMOS-compatible materials and processes borrowed from the microelectronics industry can be used to fabricate very low cost photonic sensing devices. But their practical application in developing commercial LoCs is limited by the significant cost of the readout instruments required for these chips. This is because the sensor relies on tracking the shift in the spectral response of the photonic structure, which involves either a tunable laser or an optical spectrum analyzer at a cost generally ranging from $30,000 to $60,000. Moreover, these components tend to be bulky and heavy, and thus not amenable to genuinely compact and portable readout systems. To overcome these problems, we have proposed a sensing system that indirectly tracks the spectral shift of a photonic bandgap (PBG) structure using a power-based readout technique.¹

Figure 1. Top: Scanning electron microscope image of a silicon-on-insulator 1D photonic bandgap structure used for sensing. Bottom: Close-up view of the periodic transversal elements.

PBG structures are periodic dielectric structures that show a forbidden band in their transmission spectrum The position of this forbidden band depends on the refractive index of the surrounding medium. Therefore, any target analyte binding to the surface of the sensing structure will provoke a spectral shift of its PBG position. Our indirect tracking method consists simply of exciting the PBG sensing structure with a limited bandwidth source aligned with the PBG edge, and measuring the output power. Because the output power is given by the overlap between a square-like function (the source spectrum) and a step-like function (the PBG edge from the sensor), any shift in the PBG edge position during sensing will be translated into a power variation.

Figure 2. Time evolution of the output power for anti-bovine serum albumin antibody detection experiments at different concentrations. The sensor is regenerated after each detection step. PBS: Phosphate-buffered saline. nM: Nanomolar.

We experimentally demonstrated this low-cost sensing technique using a 1D PBG structure (see Figure 1). The structure was fabricated in a silicon-on-insulator (SOI) wafer by CEA-LETI (France) in the context of the European Commission co-funded ePIXfab platform and consequently is compatible with mass-manufacturing techniques. We performed both refractive index experiments (involving different concentrations of sodium chloride in water) and antibody-sensing investigations (using various concentrations of anti-bovine serum albumin—antiBSA—in phosphate-buffered saline). Based on our findings, we estimated extremely low detection limit values, for example, in the order of 2×10⁻⁶ RIU (refractive index units) for variations in refractive index, and in the order of 10pM for antiBSA antibody detection (see Figure 2).

The main advantage of our approach compared with most typical photonic sensing methods is that, because it does not require tunable sources or detectors, the cost of the readout system is reduced from several tens of thousands US dollars to a few thousand. It also allows significant reduction in the size and weight of the final system. Moreover, the system is simpler because the sensing information is directly obtained from the measured output power and not from any spectral feature. Continuous monitoring of the output power also allows performing true real-time sensing with a very high sensitivity, which allows the instantaneous observation of any interaction of the analyte with the sensing structure.

Finally, the power-based readout technique offers an additional key feature: the ability to integrate broadband optical sources and photodetectors with the sensing structure, hence increasing the compactness of the final system, as well as enabling purely electrical access to the photonic sensing device.

In summary, we have reported a readout technique that should enable fully-integrated electro-optic sensors. Accordingly, our group is currently developing biosensors based on silicon and other CMOS-compatible materials such as silicon nitride and aluminum oxide, which are suited for working in the visible-wavelength range where source integration is a more mature process. We are also working on novel chemical bio-functionalization procedures to significantly improve the sensitivity and selectivity of our photonic biosensors.

We acknowledge funding from the European Commission under contract FP7-295043-BELERA, from the Spanish Ministry of Science and Innovation under contract TEC2008-06333, and from the Generalitat Valenciana through the PROMETEO-2010-087 R&D Excellency Program. The current work on integration of sources and photodetectors is being carried out within the FP7-BELERA project in collaboration with the Belarusian State University of Informatics and Radioelectronics.