Ultrasensitive and high-throughput fluorescence analysis of droplet contents with orthogonal line confocal excitation

Abstract

This paper describes a simple modification to traditional confocal fluorescence detection that greatly improves signal-to-noise (s/n) for the high-speed analysis of droplet streams. Rather than using the conventional epi geometry, illumination of the droplet was in the form of a line that is orthogonal to both the direction of flow and the light-collection objective. In contrast to the epi geometry where we observed high levels of scattering background from the droplets, we detected more than 10-fold less background (depending on the laser power used) when orthogonal-line-confocal illumination was used. We characterized this improvement using a standard microfluidic platform over a range of analyte concentrations and observed an improvement in limits of detection of greater than 10. Using this method, we were able to analyze picomolar concentrations of analytes contained within picoliter-volume droplets at a rate of greater than 350 droplets per second.

Figure 1

Microfluidic system for droplet formation and manipulation. (a) Schematic showing the overall chip design. The dotted circles labeled (b-e) in the schematic represent regions of the chip that are shown in the corresponding panels (b-e). In this chip design, the droplets were first formed using a T-channel geometry (b), after which the spacing between the droplets was adjusted using hydrodynamic focusing (c). The droplet stream then flow through a serpentine channel system (d) – which may be useful for different droplet manipulations such as for continuous-flow PCR or simply as a delay line to allow time for hybridization of DNA or binding of protein targets - before each droplet was interrogated at the detection region (e). (f) Photograph of a fully assembled chip with tubing and interconnects; the electrical connections (copper pads) attached to the underside of the cover glass, which was patterned with ITO electrodes, allow for on-chip heating. Scale bars in (b-d) represent 200 μm; scale bars in (e) and (f) represent 100 μm and 10 mm respectively.

Figure 2

Schematic illustration showing the two optical systems that are being compared. (a) shows the large area epi-confocal arrangement, and (b) shows the arrangement for orthogonal-line-confocal excitation. OBJ = Microscope objective, DBS = Dichroic beam splitter, PH = Pinhole, BPF = Band pass filter, AL = Aspheric lens, APD = Avalanche photo diode.

Figure 3

Optical configuration for orthogonal line confocal excitation. (a) Schematic layout of the optical train used to generate the orthogonal line excitation; T1 and T2 are telescopes, AL is an aspheric lens, and OBJ is the microscope objective. (b) Photograph showing the actual layout of the optical train; blue arrows were drawn in to illustrate the beam path. (c) An overlay image showing a droplet transiting the excitation volume; the laser excitation volume was imaged by filling the channel with fluorescein and by imaging the emitted fluorescence from the excited dyes. The scale bar in the image represents 30 μm.

Figure 4

Detection of droplet streams. (a) Time trace showing the detection of individual passing droplets (each photon spike corresponds to one droplet) using the epi-confocal geometry. Each droplet contains 5 nM of fluorescein. (b) Detection of the same droplet stream as in (a), but with orthogonal-line-confocal excitation. (c) Even at 250 pM concentrations of fluorescein, which corresponded to ~2200 molecules of fluorescein in the 30 μm diameter droplet, the signal-to-noise ratio was excellent. The laser power used in (a) and (b) was 0.75 mW, and in (c) was 7 mW.

Figure 5

Comparisons of background and signal-to-noise between epi- and orthogonal illumination strategies. (a) and (b) compare both background and noise, respectively. (c) shows a comparison between the peak heights normalized against excitation power for both illumination strategies at each concentration. The variance in the signal peak heights between both excitation methods was measured to be within a 95% confidence interval. (d) depicts the accumulated results, showing an increase in signal-to-noise ratio and the increased sensitivity through lowering of the background signal that allows for detection at much lower concentrations.