Recently, the idea of applying optical tweezers on a chip using the waveguide near field has been gaining increasing interest in optofluidics areas for lab-on-a-chip applications. The use of micro/nanoresonators to obtain enhanced optical fields and functional devices further extends the optical manipulation ability on a chip [
1
D. Erickson, X. Serey, Y. Chen, and S. Mandal, Lab Chip
11, 995 (2011).
[CrossRef] [PubMed]
]. Mandal
et al. first demonstrated the nano/microparticles manipulation on a silicon photonic crystal resonator [
2
S. Mandal, X. Serey, and D. Erickson, Nano Lett.
10, 99 (2010). [CrossRef]
]. Yang
et al. [
3] and Lin
et al. [
4] demonstrated optical switching of the microparticles on SU8 and silicon microring notch filters. Cai
et al. demonstrated micro particles add-drop devices using silicon nitride (SiN) microring add-drop filters [
5]. Arnold
et al. demonstrated nanoparticle manipulation and sizing using whispering-gallery modes (WGMs) on a silica microsphere [
6
S. Arnold, D. Keng, S. I. Shopova, S. Holler, W. Zurawsky, and F. Vollmer, Opt. Express
17, 6230 (2009).
[CrossRef] [PubMed]
].
In this Letter, we report optical manipulation of microparticles on a SiN microdisk resonator in an integrated optofluidic chip based on our previous work [
7
H. Cai and A. W. Poon, in Proceedings of Conference on Lasers and Electro-Optics (Optical Society of America, 2011), paper JWA110.
]. By tuning the laser wavelength to excite various microdisk WGMs, we demonstrate particle manipulation on a microdisk resonator with different trapping tracks and extended trapping ranges. We also demonstrate that multiple trapping tracks can be realized using single resonance wavelengths. Multiple particles can be assembled along the trapping tracks on the microdisk resonator.
Figure
1a illustrates the principle of optical manipulation of microparticles using a microdisk resonator. Unlike microring resonators which are single-mode devices, microdisk resonators usually support multiple WGMs. High-order WGMs exhibit multiple mode-field maxima (MFM) along the radial direction, which potentially form the multiple particle trapping tracks. The extended mode-field distribution inside the microresonator of the high-order mode extends the particle trapping range inward from the disk edge. The evanescent field outside the microdisk sidewall also enables an additional particle trapping track. Particles trapped by these tracks are driven by the optical force [
8
J. T. Rubin and L. I. Deych, Phys. Rev. A
84, 023844 (2011).
[CrossRef]
] of these resonance traveling waves as demonstrated in microring resonators [
3,
4,
5,
6
S. Arnold, D. Keng, S. I. Shopova, S. Holler, W. Zurawsky, and F. Vollmer, Opt. Express
17, 6230 (2009).
[CrossRef] [PubMed]
].
We fabricate
diameter,
thick SiN microdisk resonators on top of a
silica undercladding layer on a silicon wafer using
i-line photo-lithography and dry-etching. Figs.
1b,
1c show the scanning electron microscope (SEM) pictures of the coupling region between the input-coupled waveguide (with a designed width of
) and the microdisk. The coupling gap spacing is
. The silica microfluidic channels, which are also fabricated by
i-line photo-lithography and wet-etching, consist of
height silica walls and a cover glass [
5].
We inject diluted colloidal solution of
sized polystyrene particles into the fluidic channel (
). For spectra measurement, we end-fire
TM-polarized (
) laser light around
wavelength into the tapered input- waveguide using a polarization-maintaining lensed fiber. Figure
2a shows the throughput- and drop-spectra of the waveguide-coupled microdisk add-drop device with colloidal solution cladding. The spectra show two free spectral ranges containing multiple WGMs with quality factor (
Q) from
to
.
Figure
2b shows the 3D finite-element-method (FEMLAB COMSOL) simulated eigen-modes of a
diameter,
thick SiN microdisk resonator without coupled waveguides. The calculation assumes an upper-cladding of water [refractive index (RI) of 1.32 at
range] and a silica layer (RI of 1.44 at
range) as an undercladding. We tune the RI of SiN (
) to align the calculated mode positions and spacings with those of the measured spectra, assuming the resulted effective RI takes into account the perturbation of the coupled waveguides. The simulation suggests that TM (
ν,
l) modes (
ν and
l are the radial and azimuthal order numbers)
,
,
,
,
, and
are aligned with the measured resonance modes A’, B’, C’, A, B, and C, respectively, using
with
tolerance. The comparison between the measured spectra and the simulation suggests the absence of the first-order modes. Considering the surface roughness of the microdisk edge due to the dry-etching process [see Fig.
1b], first-order modes which are the closest to the microdisk edge might experience higher scattering loss than higher-order modes. The alignment of the calculated TM
and the small peak next to resonance C’ in the measured drop-spectrum also suggests that first-order modes could be suppressed. The simulation suggests no matching for resonances D’ and D.
For particle manipulation, we end-fire erbium-doped fiber amplifier (EDFA) amplified
TM-polarized laser light into the input-waveguide. The insertion loss is
. We estimate the coupling loss to be
considering the mode area mismatch between the lensed fiber and the tapered waveguide and also the waveguide end-face reflection loss. The estimated power in the input-waveguide is
near the microresonator, which is located at
from the waveguide input end-face. Figure
2c shows the zoom-in throughput-spectra at resonance
B with spectrum red-shifting and extinction ratio reducing upon increasing the laser power. We attribute this to the SiN absorption-induced thermal-optic effect at
range [
9
K. Wörhoff, P. V. Lambeck, and A. Driessen, J. Lightwave Technol.
17, 1401 (1999).
[CrossRef]
]. The multiple resonance modes upon high-power input are red-shifted to wavelengths A’ (
), B’ (
), C’ (
), D’ (
), A (
), B (
), C (
), and D (
). Images of particles are taken by a microscope system with a
objective lens and a charge-coupled device (CCD) camera.
We observe one particle travels two round trips with laser power from
to
upon resonance
B. Figure
2d shows the particle trajectories. The particle is captured from the fluidic medium at laser power of
(estimated input-waveguide power of
) at position 1 with (
R,
θ) of (
,
). The particle is temporarily stuck at position 2 (
,
) until the laser power is increased to
(estimated input-waveguide power of
). The particle starts the first round trip at position 2, travels toward position 3 (
,
) and finishes the first round trip at position 5 (
,
) after
. The particle finishes the second round trip at position 8 (
,
) after
. (
Media 1) The spikes at positions 4 and 7 in Fig.
2d indicate that the particle gets stuck temporarily and suggest that the measurement uncertainty for single-particle is
and
.
Figure
2e shows the corresponding angular velocity values of the particle. The oscillation of the velocity is probably affected by the microdisk surface roughness [see Fig.
1b] and the fluidic environment. The particle travels at an average angular velocity of
upon higher laser power (
), which is
faster than that upon the lower laser power (
).
Figures
3a,
3b,
3c,
3d,
3e,
3f show the radial position
R versus angular position
θ plots of accumulated particles trajectories (a total of
N single-particle events) upon resonance A’(
), B’(24), C’(23), A(30), B(20), and C(22). The particles travel in multiple tracks on the disk (tracks 1 and 2) with
and along the micro resonator sidewall edge (track 3) with
(indicated by the green lines, with the microresonator edge indicated by the red line). Between the trapping tracks, particles cannot be trapped steadily.
Figures
3g,
3h,
3i show the simulated cross-section of the resonance mode-field distributions of TM
,
, and
. For mode
, the inner- most and the second MFM located at
and
, are consistent with tracks 1 and 2 of resonances A’ and A. For mode
), the inner-most MFM (
) is consistent with track 1 for resonances B and B’. The second MFM (
) is consistent with track 2 for resonance B. The third MFM cannot be identified in the recorded trajectories, which we attribute to a weak surface field. For mode
, the inner-most MFM (
) is consistent with track 1 for resonance C and C’. Track 2 locates between third (
) and fourth (
) MFM could be due to the interference between resonance C (or C’) and the resonance nearby [see Fig.
2a]. Table
1 summarizes the particle trapping tracks, measured
Q values and the simulated MFM for the second–fourth order modes.
We switch the particle traveling tracks by tuning the laser wavelength from resonances B-A-C to an off- resonance wavelength with laser power of
(estimated input-waveguide power of
). Figure
4a shows the optical micrograph of microdisk with particle positions (indicated by the red circles) at various times. Figures
4b,
4c,
4d,
4e,
4f,
4g,
4h,
4i show the zoom-in images of the particle at various times. Upon resonance
B, the particle is captured from the fluidic medium at (
,
) [Fig.
4b] and travels toward (
,
) [Fig.
4c] with trapping track of
. Upon tuning to resonance
C, the particle shifts to a track with
[Fig.
4d] until it reaches (
,
) [Fig.
4e] where the particle is temporarily stuck (for
). Upon tuning to resonance
A, the particle shifts its track again (
) [Fig.
4f]. When the particle travels back to (
,
) [Fig.
4g], the laser wavelength is tuned to an off-resonance wavelength at
, then the particle is trapped by the input- waveguide [Fig.
4h] and coupled out from the throughput-waveguide [Fig.
4i]. Figure
4j shows the particle trajectory in an
R versus
θ plot with various trapping tracks indicated by the green lines. Figure
4k shows the particle angular velocity values during the laser wave length tuning process.
We also observe multiple particles are assembled along the trapping tracks on the microresonator. Upon laser power of , we record 20, 20, and 13 particles assembling upon resonances A, B, and C, respectively.
In summary, we demonstrated optical manipulation of microparticles on a SiN microdisk resonator in an integrated optofluidic chip. We observed up to three trapping tracks within one high-order WGM and extended trapping range up to
inward from the resonator edge. We demonstrated various sets of trapping tracks and extended trapping ranges through various high-order WGMs. We also observed that particles can be assembled along the tracking tracks on the microresonator. This multimode resonator with multiple trapping tracks, extended trapping ranges, and potentially being a particle assembler, could be a potential building block for our previously proposed particle-circuits [
5].