Amorphous Silicon Based Biosensor For Milk’s Fat Detection

Lab-on-Chip (LoC) systems perform complex biomolecular analysis in several fields such as diseases diagnosis, food quality control, tissue engineering and pharmaceutics [1] integrating optics, chemistry, microfluidics and micro-nano technologies. When optical techniques are employed, complex and bulky optical setups are usually needed to deal with off-chip detection. We present a compact and cost-effective optoelectronic LoC able to detect a given analyte’s presence and concentration in a biological sample [2]. In this system, the sample under investigation is brought in direct contact with a waveguiding structure and interacts with the light’s evanescent wave; the extent of the interaction is measured by an embedded photodetector, providing information on the mixture’s composition. Fig. 1 depicts the structure of the proposed system: a SU8 polymer optical channel was chosen since is a simple and fast-prototyping waveguide on a glass substrate; a n-doped/intrinsic/p-doped hydrogenated amorphous silicon (a-Si:H) photodiode was employed as versatile, cost-effective detector, taking advantage of its low dark current (about 100 pA/cm2) and high responsivity in the visible spectrum (up to 500 mA/W). We report a demonstration study on fat detection in milk. The analysis was performed by interpolating the output data from simulated interactions between an optical waveguide and milk samples (whose optical properties at different fat concentrations were found in literature [3]), with the measured electro-optical response of the system. COMSOL Multiphysics® was used to model a section of the interaction site (see Fig. 2a) and to evaluate the amount of waveguided optical power that enters (“in” box, see Fig. 2b right) and leaves (“out” box, see Fig. 2b left) the site for 5 different samples. All the simulations have been conducted considering light excitation at the wavelength of 660 nm. Fig. 3a shows the percentage of absorbed optical power with respect to the 5 different milk samples. Assuming a detector’s photoresponse of about 125 mA/W at 660 nm and a traveling light entering the device with an optical power of 40 nW before interacting with the sample, Fig. 3b plots the estimated photocurrents resulting as a consequence of light’s interaction with the 5 modelled samples. The sensor’s photocurrent shows a linear trend with a span ΔI equal to 287 pA corresponding to a fat concentration span Δc of 0.33. The system’s sensitivity to the fat content variation (ΔI/Δc) is about 8.7 pA/(g/dL). The Schottky noise current fluctuation in this state was calculated to be around 30.9 fA, so a minimum detecting signal (MDS) equal to three times the noise, 92.6 fA, is defined. Finally, the limit of detection of our system was determined as MDS·(Δc/ΔI), and is approximately 106 ppm. These results display promising performances and encourage further efforts toward an actual implementation in food-quality applications.