Droplet-based Biosensing for Lab-on-a-Chip, Open Microfluidics Platforms

Abstract

Low cost, portable sensors can transform health care by bringing easily available diagnostic devices to low and middle income population, particularly in developing countries. Sample preparation, analyte handling and labeling are primary cost concerns for traditional lab-based diagnostic systems. Lab-on-a-chip (LoC) platforms based on droplet-based microfluidics promise to integrate and automate these complex and expensive laboratory procedures onto a single chip; the cost will be further reduced if label-free biosensors could be integrated onto the LoC platforms. Here, we review some recent developments of label-free, droplet-based biosensors, compatible with "open" digital microfluidic systems. These low-cost droplet-based biosensors overcome some of the fundamental limitations of the classical sensors, enabling timely diagnosis. We identify the key challenges that must be addressed to make these sensors commercially viable and summarize a number of promising research directions.

Keywords: biosensors; droplet; early detection; high sensitivity; lab-on-a-chip; point-of-care.

Figure 1

A droplet-based LoC platform must be integrated with highly sensitive and selective sensors. (a) General configuration of digital microfluidics platforms. Digital microfluidics offers a broad range of droplet operations (e.g., generation, transport, mixing, sensing, etc.). This review focusses on droplet-based sensors and their performance limits. (b) In a closed microfluidic system, sensors analyze the droplets as they flow past the sensors; (c) In an open microfluidic system, the droplet is placed on the sensor surface, and no continuous flow is required. Figure 1(c1–c4) show various aspects of droplet-based sensors covered in this article.

Figure 2

(a) Left: On a symmetric surface, a droplet forms a semispherical cap-shaped structure with a circular contact line. Right: It forms an oval-shaped contact line on an asymmetric surface, such as the structure in Reference [11]. The inset shows computer graphic of a lotus leaf surface; (b) SEM image of the electroplated electrodes. The figure on the right shows an AFM profile of the electrodes’ nanotextured surface; (c) An optical image of a droplet on the electrode array 4 min after deposition; (d) The same droplet 10 min later; (e) The relative conductance change as a function of the initial DNA concentration. Figures are reproduced from Reference [67] by permission of the Royal Society of Chemistry. Inset of Figure 2a is reprinted with permission from @ William Thielicke.

Figure 3

(a) Schematic of a FET/nanowire biosensor with on-chip electrodes for localized desalting and simultaneous device biasing. Positive $\left(A^{+}\right)$and negative $\left(B^{-}\right)$ ions are attracted towards negative and positive polarity electrodes, respectively, depleting the droplet bulk of salt; (b) Numerical calculation of ion profile showing negative ion density in a 300 pL droplet (6100 $\mu m^2$ electrode area) at $1\mu M$ (background strength under 1 V desalting bias); (c) Ratio of the droplet- volume to the electrode-area required for desalting the droplet by 50%, as a function of desalting voltage and ionic concentration. For example, desalting at 100 mM concentration under 1 V desalting bias requires an aspect ratio of ~1 $\mu m$ . Reproduced with permission from Appl. Phys. Lett. 106, 053105 (2015). Copyright 2015, AIP Publishing LLC.

Figure 4

(a) Schematic of a droplet sitting on top of a FET device; (b) An array of droplets sitting on linked devices for parallel detection; (c) Simulated temperature profile within the droplet for an applied bias 36 V. Temperature within the droplet is highly localized, and returns close to room temperature at the edges minimizing the evaporation; (d) Theoretical estimate of the droplet temperature as a function of applied ac bias. Temperature varies roughly as a square of the applied AC bias. (Figure 4b–c are adapted and Figure 4d replotted from Reference [67] with permission from National Academy of Sciences).

Figure 5

Derivative of fluorescence w.r.t. voltage vs. AC voltage for 3 DNA strands, the red and black curves correspond to DNA samples with fully-complementary strands and the blue curve a hetroduplex with a single-base pair mismatch. The hetroduplex showed the peak at lower voltage, thereby indicating a single-base pair mismatch (because of lower melting temperature). Figure replotted from Reference [67] with permission from National Academy of Sciences.

Figure 6

(a) State-machine shows how various solution evolve as their binding state changes through time. Solid, dashed, and dotted-dashed lines represent $S_{\pi }$ (full match), $S_Y$ (partly-match), and $S_{\left|\right|}$ (full-mismatch) solutions, respectively; (b) Plot of the first principal component obtained from (i) a data set comprised of the results of the initial state and 1st incubation (total evaluation time of 80 min), and (ii) by considering the results obtained from the 1st heating step to the data set (total time ~85 min). Selective detection down to 2 nM is realized after ~85 min.