Free-Space Excitation of Optofluidic Devices for Pattern-Based Single Particle Detection

Abstract

Optofluidic sensors have enabled single molecule sensing using planar, waveguide dependent multi-spot fluorescence excitation. Here, we demonstrate a new approach to single-particle fluorescence sensing using free-space, top-down illumination of liquid-core antiresonant reflecting optical waveguide (ARROW) devices using two different multi-spot excitation techniques. First, the liquid core ARROW waveguide is excited with a focused beam through a slit pattern milled into an opaque aluminum film, showing comparable performance for single bead fluorescence detection as in-plane, multi-mode interference waveguide based excitation. The second top-down illumination technique images the spot pattern from a Y-splitter SiO₂ waveguide chip directly onto the detection device for efficient power utilization and circumventing the need for an opaque cover, producing a further 2.7x improvement in signal-to-noise ratio. The two top-down approaches open up new possibilities for chip-based optical particle sensing with relaxed alignment tolerances.

I. Introduction

The need for rapidly identifying a potential disease from biological specimens using compact, high-precision technology has been a key focus in the recent diagnostics research [1], [2]. Different integrated photonic technologies have shown great potential in point-of-care biosensing by implementing a variety of sensitive schemes in lab-on-chip platforms [3]–[6]. In particular, antiresonant reflecting optical waveguide (ARROW) based optofluidic sensors have shown strong capabilities to detect a broad range of target biomarkers with single-particle sensitivity, e.g. DNA, pathogenic protein, and single virus by leveraging optical signal guiding in a low-index liquid medium [7]–[10]. Multi-spot excitation of moving particles enables their detection with high signal-to-noise ratio [11]. This principle was implemented on the ARROW platform using multi-mode interference waveguide (MMI WG) or Y-splitter waveguides [12] and multi-mode interference (MMI) waveguides [13], with the latter enabling spatial and spectrally multiplexed detection on the single biomolecule level [10], [14].

Waveguide-based planar excitation also comes with the challenge of efficiently coupling light into the excitation waveguides. This is typically accomplished using in-plane coupling with a butt-coupled fiber or out-of-plane excitation with a grating coupler [15]. However, both approaches require precise alignment, either to the waveguide mode or the grating resonance (angle). Moreover, implementing spectral multiplexing in the MMI waveguide-based ARROW devices places additional constraints on the waveguide dimensions, increasing overall fabrication complexity [16]. MMI waveguides also limit spectral multiplexing to multi-spot excitation patterns with equidistant (periodic) spots.

Free-space, non-planar excitation by directly shining light on the liquid core ARROW waveguide (LC-ARROW WG) offers an attractive solution to these challenges by dramatically relaxing the alignment tolerances. Non-planar schemes have been relevant for bulk measurement in plasmonic biosensors, in microfluidic devices to modulate the emission signals, and for multiplexed detection of spectrally overlapping signals [17]–[19]. In this Letter, we introduce two high signal-to-noise ratio techniques for leveraging free-space, multi-spot excitation in ARROW devices with a top-down approach. In the first approach, the LC-ARROW channel is covered with a thin, non-transparent aluminum (Al) film- into which μm-scale slits are etched to enable patterned, multi-spot illumination using a focused, direct laser beam. We investigate the fluorescence excitation capability of this scheme and compare the signals with planar, MMI waveguide based excitation. Also, spectrally exclusive multiplexed detection ability is demonstrated using spatially encoded excitation slit patterns to identify particles in a mixed sample environment. The second technique circumvents the need for the Al coating by imaging a patterned beam into multiple confined spots using Y-splitter waveguides. Aside from simplifying the fabrication process, this approach also provides flexibility to excite targets anywhere in the detection channel, offers more efficient delivery of excitation power, and improves the signal quality due to highly confined excitation spots.

II. Design, Fabrication, and Experiment

The LC-ARROW waveguide platform operation is based on antiresonant, low-loss guiding of fluorescence signals in a 12 μm × 5 μm liquid channel placed over periodic dielectric layers stacked on a silicon (Si) substrate, before ultimately coupling into a solid core collection waveguide. Solid-core excitation ARROWs were fabricated either with single oxide ridge waveguides [8], [20] (here used for the Y-splitter devices) or with a 6 μm thick high refractive index (1.51) core SiO₂ layer buried into another 6 μm thick protective cladding SiO₂ layer with low refractive index (1.448) [21] (here used for MMI waveguide devices). In particular, the LC-ARROW channel was created using an SU-8 sacrificial core and a wet etching step. A target particle inside the LC-ARROW channel can be excited by either type of these orthogonally intersecting waveguides (Fig. 1(a)). For top-down, direct laser excitation, the same buried structure as in planar, MMI waveguide excitation is utilized by covering the top surface with a 100 nm thin reflecting Al film, which was deposited using e-beam evaporation at a slow rate (<0.05 nm/s) to avoid flaking off. Finally, 4 μm × 15 μm, 100 nm deep light entrance slits with different patterns were milled using gallium (Ga⁺) focused ion beam (FEI Quanta 3D) with an experimentally set 30 kV, 0.5 nA nominal current and 550 nm nominal depth following its Si milling protocol (dwell time-1 μs, volume/dose- 0.15 × 10⁻⁹ m³/s). Fig. 1(b) shows an SEM image of a 7 slit excitation pattern. Each pattern’s dimensions and spacing can be reconfigured based on the application needs.

Fig. 1.

(a) Schematic of ARROW optofluidic device with planar waveguides and top-down slit patterns indicated, (b) Top view SEM image of the metal coated ARROW device with 7 slit pattern, (c) experimental setup for fluorescence excitation and detection, and (d) example of detected events (circled in red) by the APD.

Fig. 1(c) shows the experimental setup used for excitation and collection of fluorescence signals from the ARROW devices. A 19 mW, 635 nm laser beam is incident on the slits after passing through a 90:10 beam splitter and a 10× objective, which also redirect the chip surface reflected image towards an alignment camera. An additional 556 nm laser beam was sent through the beam splitter during the multiplexed detection using a dichroic mirror. For patterned excitation on an uncoated device, the laser source was replaced with the fiber coupled Y-split waveguide followed by a 10×, long working distance collimation objective. Fluorescence signals from particles in a negative pressure driven fluidic flow are collected, filtered, and finally detected by an avalanche photodetector (APD). Fig. 1(d) shows an example of data collected by the APD, each spike representing a single particle event.

III. Particle Detection and Performance Analysis

A. Comparison Between Top-Down, Al-Coated and Planar, MMI Waveguide Excitations

In order to characterize the sensor performance using focused laser beam excitation, comparative analyses were performed between the top-down and MMI waveguide excitation schemes by exciting 1 μm and 200 nm diameter (both FluoSpheres^™, 625/645 crimson) fluorescent polystyrene beads at 10⁶ mL⁻¹ concentration inside the same device. As shown in Fig. 2(a), a 7 slit pattern with 11 μm slit-to-slit spacing was milled and excited at exact alignment with the MMI excitation waveguide on the LC-ARROW channel. A 75 μm wide MMI waveguide excited by laser light (633 nm, 5.3 mW) projects a very similar excitation pattern, as observed in Fig. 2(b) by filling the LC-ARROW channel with quantum dot solution. Table 1 compares the salient optical parameters between the two excitation schemes. When the average intensity levels delivered to the LC-ARROW channel by the two schemes are matched, the top-down excitation can accommodate adjustable power levels inside the channel by varying the slit sizes, allowing more control over signal temporal properties. While a planar excitation approach requires precise alignment and mode matching between a fiber and the excitation waveguide core to generate high-fidelity spots, a free-space excitation over an Al-coated pattern provides similar multi-spot excitation volumes inside the LC-ARROW channel, with greater alignment tolerance.

Fig. 2.

(a) Attenuated top-down excitation beam incident on the slits pattern, (b) top view of the excitation spots inside the LC-ARROW channel from MMI waveguide excitation (colored image), (c)–(d) zoomed in signals from 1 μm beads using top-down and planar excitation, respectively, and (e)–(f) signals from 200 nm beads using top-down and planar excitation, respectively, and (g)-(h) signal distributions of top-down (TD) and MMI waveguide (MMI) excitation using 1 μm and 200 nm beads, respectively.

TABLE I.

Comparison of Excitation Properties and Detected Signals

Property	Top-down excitation-Al coated ARROW	Planar, MMI waveguide excitation	Top-down patterned excitation-uncoated ARROW
Excitation area (μm2)	48	11.54	38
Excitation volume (μm3)	240	138.5	189
Average excitation intensity (μW/μm2)	11.54	11.31	5.28
Incident power/ excitation spot (μW)	553	151	217
Excitation/ Source power (%)a	19.76	19.9	17.3
Signal mean ± standard deviation, 1 μm beads (counts/0.1 ms)	2906 ± 295	2816 ± 209	2016 ± 246
Signal/Noise, 1 μm beads	181.6	281.6	504
Signal mean ± standard deviation, 200 nm beads (counts/0.1 ms)	195 ± 158	161 ± 156	92 ± 92
Signal/Noise, 200 nm beads	12.2	16.1	23

^aFor excitation of an Al-coated chip, the source is the diode laser; for planar MMI excitation and for top-down patterned excitation the source is the fiber coupled to the corresponding waveguides.

Figs. 2(c) and 2(d) show the fluorescence signals emitted from a 1 μm bead using top-down and planar excitation, respectively, and Figs. 2(e) and 2(f) show the 200 nm bead signals. In each case, distinct 7-peak signals were observed corresponding to the excitation pattern. Signals from smaller beads appear sharper in both cases due to better spatial resolution. The signal level distributions are very similar for the two schemes as shown in Figs 2(g) and 2(h), with the mean and standard deviations listed in Table 1. The signal-to-noise ratios (SNR), however, are lower for top-down excitation because of higher noise level. SNR can be improved by coating the devices with thicker Al layer using improved thin film deposition processes and patterning the slits [22].

B. Multiplexed Single-Particle Detection

Multiplexed, simultaneous detection of multiple particles in a small sample volume is a key element in on-chip bioanalysis. Using top-down excitation, we demonstrate this capability in ARROW devices by milling two different slit patterns 120 μm apart on the same LC-ARROW channel (Fig. 3(a)). Pattern 1 is a periodic, 5 slit pattern with 10 μm slit-spacing, while pattern 2 consists of 3 slits with varying slit-spacing, i.e., one slit spacing is doubled (20 μm) to implement a non-periodic modulation. As shown in Fig. 3(b), patterns 1 and 2 are excited with two spatially separated focused laser beams at 635 nm (R) and 556 nm (G), respectively to identify distinctly labeled particles. Upon simultaneous (R+G) and separate laser excitations (R, G), a 10⁶ mL⁻¹ mixture of 1 μm fluorescent beads (Bangs Laboratories, 660/690 Flash Red (FR) and FluoSpheres^™, 540/560 Orange (O)) emits signals with differing amplitude levels, mostly owing to different fluorescence properties of the beads (Fig. 3(c)). A closer look at the individual signals reveals their distinct multi-peak signatures corresponding to respective excitation patterns, as shown in Figs. 3(d) and 3(e).

Fig. 3.

(a) Illustration of 5 slits and 3 slits encoded patterns, (b) attenuated 635 nm (R) and 556 nm (G) beam excitation on 5 slits and 3 slits patterns respectively, (c) multiplexed fluorescence signals from simultaneous and separate excitation with two beams, separated by vertical lines, and (d)–(e) single particle signals respectively from O and FR beads with corresponding excitation patterns.

To identify the signals based on their temporal modulations, a multiscale signal processing algorithm based on continuous wavelet transform (CWT) was implemented [23]. Briefly, custom defined wavelets for each pattern with a range of timescales Δt were correlated with the entire trace to calculate the CWT coefficients at every location. The maximum value of the CWT coefficient identifies the pattern and location of a signal. Table 2 lists the CWT-based events statistics obtained from separate as well as mixed flow of the particles from simultaneous R+G excitation. As expected from the separated bead flows, exclusive laser response with correct signal pattern from a particle is observed, which is also verified for single color (R, G) excitations. The statistics for the mixed particle flow can also be verified using amplitude-based statistics. As the signal levels from two beads can be clearly separated by the horizontal line in Fig. 3(c), an amplitude-based classification of the signals produces the same results. Such an encoded excitation scheme can be very useful as it enables multiplexed detection using fewer number of slits on a compact platform.

TABLE II.

Detection Statistics Using Simultaneous R+G Excitation

Sample beads	Detected 5 peaks events / mL(× 106)	Detected 3 peaks events / mL(× 106)
FR	1.08	0
O	0	3.3
FR + O	1.10	3.35

C. Patterned Beam Excitation on Uncoated ARROW Device

A second way to achieve free-space excitation on the LC-ARROW waveguides is to directly image an excitation pattern onto the detection channel. This eliminates the need for depositing and patterning an Al layer and is more power-efficient as the full incident power is contained in the multi-spot pattern. Here, we use a 1 × 4 Y-splitter waveguide chip (Fig. 4(a)) and focus its output pattern on an uncoated ARROW device. As shown in Fig. 4(b), the same spots were imaged using a 10× objective onto a small length of an LC-ARROW channel for multi-spot excitation.

Fig. 4.

(a) Confined spot patterns emitting from a 1 × 4 Y-split waveguide facet. Inset shows an illustration of the excitation device, (b) attenuated patterned excitation beam incident on an uncoated ARROW chip, and (c)–(d) zoomed in signals from 1 μm and 200 nm beads respectively using patterned excitation.

The same samples as in section III.A were used for characterizing the sensing performance, with results incorporated in Table 1. The reduced excitation intensity due to optical instruments loss causes lower signal levels compared to the other two approaches. Nevertheless, the highly confined excitation volume enhances the SNR with this approach than direct laser beam excitation. Better excitation confinement also results in more distinct peaks in each signal, as shown in the single signal views in Figs. 4(c) and 4(d) with 1 μm and 200 nm beads, respectively. With optimized excitation device design, the beam-patterning scheme can also be useful for different multiplexed detection strategies in optofluidic sensors.

IV. Conclusion

The two free-space, top-down excitation techniques presented here show comparable detection performances with a planar, MMI waveguide based excitation approach. Additionally, they offer the ability to transfer a non-periodic spatial encoding to the fluorescence signals to leverage the full advantage of different signal processing techniques. The multiplexing capability shows special promises for differential screening of disease biomarkers. On the other hand, beam patterning using waveguides can be an efficient, all-photonic approach for free-space, multi-spot excitation without complex optical components. In the future, dynamic reconfigurability of the excitation beam can be added, e.g. by use of a spatial light modulator. The results show exciting, new possibilities for ARROW optofluidic platforms in diagnostics and other sensing applications.