Fabrication of High-Quality Whispering Gallery Mode Microbubble Resonators

Here, we demonstrate a robust and standardized protocol for fabricating high-quality factor (Q-factor) Whispering Gallery Mode (WGM) Microbubble resonators (MBRs) with a precision glass processing machine (PGP).

We demonstrate a robust and standardized method for the fabrication of high-quality factor (Q-factor) Whispering Gallery Mode (WGM) Microbubble resonators (MBRs) with a precision glass processing machine (PGP). Microbubble resonators are a unique class of WGM devices with integrated fluidic channels, making them ideal for diverse sensing applications. Herein, we show a standardized protocol to fabricate high-Q microbubble resonators through the optimization of key performance metrics, such as Q-factor and wall thickness. We also show methods to improve the sensitivity of the platform to refractive index changes and other sensing targets through Hydrofluoric acid (HF) wet etching. Lastly, a brief analysis of the resistance of microbubbles to fluid flow is discussed, showing that smaller-diameter microbubbles exhibit greater resistance to flow for analyte delivery - a factor that should be considered for analyte delivery. The implementation of this refined fabrication protocol not only increases the success rate of device production but also reduces fabrication time. Moreover, the protocol can be expanded to other techniques used to produce MBRs, such as CO₂ laser-based methods.

Whispering Gallery Mode (WGM) microresonators are a class of optical sensors that have demonstrated enormous potential not only for the detection of single molecules and nano-particles^1,2,3,4,5,6 but also for sensing a wide range of physical phenomena such as magnetic⁷ and electric fields⁸, temperature⁹, and ultrasonic waves^10,11. Under optical resonance conditions, light is trapped within the device, leading to a significant power amplification^12,13. Any localized change to the resonator (such as the binding of a biomolecule or changes in the refractive index of the surrounding media) induces changes in the local optical environment, therefore shifting the resonant frequency or wavelength. By monitoring the shifts in resonance wavelength or frequency, one can detect and characterize analytes in real time.

WGM microresonators can be designed in a variety of geometries. Common geometries include but are not limited to, microtoroids¹⁴, microrings¹⁵, and microbubble¹⁶ resonators (MBR). Here, we focus on MBRs due to their great potential in optofluidic sensing applications. A key advantage of MBRs is their fluidic integration^17,18,19,20, which is enabled by the fabrication of the device from a microcapillary. In this design, the inline capillary facilitates the easy delivery of small volumes (i.e., microliters) of analytes in solution to the sensing area without the need for external fluidic channels, as shown in Figure 1. With their unique fluidic handling capabilities, MBRs are well suited for a wide range of sensing applications that are not easily achievable with other WGM platforms. For example, MBRs have been filled with magnetic fluids, thereby imbuing sensitivity to external magnetic fields²¹. Additionally, MBRs have also been used to control the specific orientation of gold nanorods in solution through optical torques²².

The fabrication of MBRs can be summarized as follows: Aerostatic pressure is applied inside the capillary while a small area of the capillary is locally heated. The combination of localized heating and internal pressure inflates the heated section into a spherical geometry capable of supporting high-Q WGMs, as illustrated in Figure 2. Various methods can be employed to achieve localized heating of the capillary, such as using a CO₂ laser²³, a fiber optic splicer²⁴, a hydrogen flame source²⁵, and a precision glass processing machine (PGP). The methods presented here can be expanded to other heating sources, including a CO₂ laser. The PGP is similar to an optical fiber splicer but offers enhanced control over heating time, power setting, and the positioning of fibers or capillaries²⁶. PGPs often include built-in microscopes adjacent to the heating elements, enabling real-time monitoring of the fabrication process. Typically, light from a tunable diode laser is coupled into the MBR via a tapered optical fiber that is in contact with the equator of the MBR. The fiber is tapered (to ~1 μm) to enable efficient coupling of light into and out of the MBR. The resulting transmission spectra from the MBR are then captured by a photodetector through the optical fiber and visualized on an oscilloscope.

Sensing with WGM MBRs relies on the interaction of the WGM field with the target analyte. The strength of this interaction is directly proportional to the fraction of the WGM field that penetrates the MBR's hollow cavity where liquid or gas phase samples can flow through²⁷. As shown in Figure 3, COMSOL simulations illustrate how the penetration of the WGM field into the inner cavity varies with the MBR's wall thickness. Maximum field penetration of the WGM field occurs as the wall thickness is reduced to less than 1 μm, with these simulations conducted using light in the 780 nm band. Achieving such reduced wall thickness through the standard heat-and-inflate fabrication protocol alone is challenging. To further thin the walls of the MBR and to make the device more sensitive, we incorporate additional wet etching steps using Hydrofluoric (HF) acid.

Using a PGP, we will focus on the fabrication of MBRs in line with a silica capillary. A detailed description of the fabrication process and methods to enhance sensitivity to refractive index changes through wet etching will also be presented.

1. Microbubble fabrication

1. Start with a polymer-coated silica glass capillary (250 μm inner diameter and 360 μm outer diameter) that is ~75 cm in length. The length of the capillary can vary by user needs; ensure that the pressure described below is reached at longer capillary lengths.
2. Burn off ~2.5 cm of the polymer coating at one end of the capillary with a butane torch and clean the end with a delicate task wipe and isopropyl alcohol (IPA).
3. Place the clean end of the capillary in the PGP and click the splice only button on the software to heat for 5 s at 180 W to seal the end of the capillary. Set the heating time and duration with a right-click on the splice only button.
4. To ensure a proper seal, examine the sealed capillary end, Figure 1B.
5. At ~25 cm from the sealed end (or 1/3 the capillary length from the sealed end), burn off a ~2.5 cm long strip of polymer coating from the capillary and clean the resulting area with IPA.
   NOTE: The position described above is not critical as pressure is evenly distributed along the capillary length. If the user's experimental setup warrants a different placement of the MBR, this should not affect the rest of the procedure.
6. Repeat this step until no polymer is visible under the imaging system, which displays capillary image with sub-micron resolution.
7. Place this newly cleaned section of the capillary in the glass processor instrument over the heating element and under the microscope described earlier.
8. Using a gas-tight syringe and a syringe pump, inject air into the capillary such that the internal pressure reaches 10 bar (i.e., decreasing the volume in a syringe from 5 mL to 0.5 mL, with corresponding pressures calculated with Boyle's Law). Fit the capillary to a syringe with a luer lock to a 360 μm adapter. Alternatively, apply pressure via a pressure regulator and compressed air/inert gas.
9. Heat the capillary with the PGP using a filament power of ~100-110 W for 1.5-2 s.
10. Conduct additional heating steps with 90-100 W and 0.1-1 s firings of the filament. This method will slowly and controllably increase the MBR's diameter to the desired range. However, this process may result in asymmetric MBRs. If an asymmetric MBR is observed (the sphere is not symmetric about the capillary axis), rotate it about the capillary axis between successive heating stepsto promote symmetry.
11. Inspect the MBR under a microscope for quality control, looking for things such as dust, cracks, or deformations. Additionally, use the microscope here to estimate the wall thickness of the MBR.

2. Wet etching with hydrofluoric acid

CAUTION: Hydrofluoric acid is very dangerous, toxic, and corrosive. Calcium gluconate should be kept nearby as this chemical can neutralize hydrofluoric acid. Wear appropriate personal protective equipment and follow all safety precautions in the Material Safety Data Sheet (MSDS).

1. Remove the sealed end of the capillary by cutting 2 cm below the sealed end.
2. Place one capillary end inside a blunt tip to the luer lock adapter. Make sure the blunt tip's inner diameter is as close as possible to the outer diameter of the capillary while making sure the capillary can still fit inside.
   NOTE: The adapter material must not be reactive with HF acid.
3. Apply UV curable epoxy to the end of the blunt tip with the capillary inside to join the two, then cure with UV light (50-100 W UV light source; higher wattage will speed up the curing).
   CAUTION: Appropriate eye protection must be worn when operating a UV light source²⁸.
4. Once the epoxy is cured, attach the luer lock tip to a 25 mL syringe that is inert to HF.
5. Place the syringe in a syringe pump with a withdrawal rate of 50 μL/min.
6. With one end of the capillary connected to the syringe, place the other end of the capillary into deionized (DI) water.
7. Set the syringe pump to withdraw water through the capillary into the syringe to check that proper flow has been established (i.e., the system is free from air bubbles with liquid consistently flowing through the capillary).
8. Transfer the capillary end from the DI water to a container of HF acid once proper flow is established.
9. Calculate an approximate etching time based on the MBR wall thickness before etching, considering a measured etching rate of 8.18 μm/h with 12% (w/v) hydrofluoric acid. Wall thickness can be measured under an imaging system that displays MBRs with sub-micron resolution.
   NOTE: The etching rate for acids of other concentrations can be empirically determined. This can be done by measuring the capillary wall thickness before and after a set etching time.
10. Run the syringe pump at 50 μL/min for the calculated etching time.
11. Rinse the capillary with DI water for 10-15 min to remove all acid from the capillary after the etching time has elapsed. Calculate the rinsing time to rinse the entire capillary volume 3-5 times.
12. Measure the MBR wall thickness after etching. Use this information to update the etching rate for the acid.
13. Repeat this process can be repeated until the desired wall thickness is achieved.

A representative MBR fabricated with the PGP machine is shown in Figure 1C. Given our starting capillary outer diameter (OD) of 360 μm, we expand the capillary ~2x in the fabrication process. Expanding the capillary to ~700 μm results in wall thicknesses between 5 μm and 15 μm. It has been shown that the optimal wall thickness for biosensing with MBRs is on the order of the wavelength of light used to excite the WGM²⁷. MBRs can theoretically achieve a quality factor of 1 x 10⁹, but 1 x 10⁶ is sufficient for most biosensing applications^29,30,31.

To validate the simulated results in Figure 3, we evaluated the response of MBRs with varying wall thicknesses to various concentrations of sodium chloride solutions. Figure 4 confirms the simulated results, showing a significant increase in refractive index sensitivity when the MBR wall thickness was around 1 μm. The three "thick-walled" MBRs (i.e., wall thicknesses of 9.4 μm, 7.4 μm, and 5.0 μm) showed a diminishing response to the refractive index changes as the wall thickness increased, as expected. Figure 5 presents the typical transmitted spectrum of an MBR. Over a broad scanning range, the spectrum exhibits high modal density. Within a narrow scanning range of 40 pm, the laser scans across multiple resonances. By tracking the resonance shift, one high-Q mode can be selected for sensing within this fine scanning range. Quality control metrics for MBR fabrication can be used to qualitatively assess both physical and optical properties to optimize biosensing performance. Two important metrics include the quality factor of the resonator (≥1 x 10⁶) and a small wall thickness for maximum interaction between the WGM and the target analyte (<1 μm).

As previously mentioned, the integrated optical and fluidic handling is an inherent strength of MBRs, making them appealing for sensing applications. Given this, we sought to explore the fluidic properties of the capillaries. By experimenting with capillaries of varying lengths20 cm, 40 cm, or 80 cm and testing different flow rates, 100 μL/min, 250 μL/min, and 500 μL/min, for each capillary length, we found that the inner diameter (ID) of the MBRs significantly influences the fluid transport efficiency (Table 1). Specifically, as the MBR ID increases from 75 μm to 250 μm, the efficiency of liquid transport through these microcapillaries significantly improves, allowing 95-100% of the set volume of liquid to be pulled through the microcapillaries. This enhanced capability highlights the potential of larger-diameter MBRs in optimizing fluid handling, making them particularly suitable for diverse sensing applications where fluid dynamics are crucial.

Figure 1: Overview of experimental setup. (A) Schematic of a waveguide-coupled WGM MBR sensor system. The WGM is excited by a tunable diode laser and monitored with a photodetector through a tapered optical fiber waveguide. The analyte is then delivered through the inline fluidic channel where it interacts with the WGM field. (B) Micrograph of a 360 μm capillary sealed by a PGP. (C) Micrograph of an MBR after fabrication. Arrows indicate where two micrographs were merged. Please click here to view a larger version of this figure.

Figure 2: Schematic of the fabrication process for MBRs, including the optional HF etching step. (1) Start with a silica glass capillary (360 μm OD). (2) Remove the polymer coating with a small flame and clean the surface with Isopropyl Alcohol (IPA). (3) To build up internal pressure, one end of the capillary must be sealed. This step is done with the PGP. (4) Increase internal pressure with air and fire the PGP filament to locally heat the capillary and inflate an MBR into a spherical geometry. (5) Optional. Etch the interior of the capillary with HF acid to thin the walls and increase sensitivity in biosensing. Please click here to view a larger version of this figure.

Figure 3: WGMs field penetration simulations. (A) Simulation of WGM penetration into an MBR when the wall thickness is in the submicron range. (B-D) Electric field intensity within the MBR for different wall thicknesses, showing greater penetration of the evanescent field into the core of the resonator for walls that are <1 μm thick. Please click here to view a larger version of this figure.

Figure 4: MBRs response to varying RI. Response of MBRs of different wall thicknesses to varying concentrations of NaCl solutions, showing a notable improvement when the wall enters the 1 μm regime due to greater WGM penetration into the liquid flowing through the inside of the resonator connected to a fluidic channel in the capillary structure. Please click here to view a larger version of this figure.

Figure 5: A typical WGM spectrum for an MBR. (A) Large scanning range over one free spectral range. (B) Fine scanning across a few WGMs. (C) Zoom-in view of a WGM in B) and its curve fitting showing a Lorentzian lineshape. Please click here to view a larger version of this figure.

Table 1: Percentages of DI water pulled through two different-sized silica capillaries.

Here, we described the protocol to fabricate high-quality whispering gallery mode (WGM) microbubble resonators (MBRs) using a precision glass processor. We present critical steps in the fabrication protocol, including the heat and expand steps. Here, a combination of overheating, heating too long, or injecting too much internal air pressure can lead to unsuccessful fabrication. To address these issues, adjustments such as lowering the heating power or heating duration in the software user interface of the PGP machine can help. However, this is not the only method used to fabricate MBRs. Several other protocols exist in the literature, but most methods share the same basic steps - heat and expand. Other fabrication methods utilize various heating sources, such as arc discharge from a fusion splicer or a CO₂ laser system, while the precision glass processor uses a graphite heating element. The arc discharge method³² is similar to the PGP in that both approaches have built-in microscopes to monitor the heating process. One major drawback of the arc discharge method is that these devices provide little control over the position of the microcapillary, making precise adjustments to the capillary position challenging.

Using a CO₂ laser³³ as a heating element offers a few advantages. In this configuration, two counter-propagating CO₂ laser beams of equal power converge on the microcapillary from opposite directions to uniformly heat the microcapillary. This uniform heating, along with internal pressure, allows for creating a symmetric MBR without the need to rotate the microcapillary during fabrication. But, operating a free-space high-powered laser has its own safety concerns and should be performed with proper training and rigorous precautions specific to the use of the CO₂ laser.

We also present methods to decrease the wall thickness of the MBRs with HF acid to improve the devices' sensitivity to refractive index changes. Using HF acid is a common method to gradually etch and thin the silica walls of the MBR, but this wet etching method can increase surface roughness and, subsequently, reduce the quality factor. Others have achieved a thin-walled MBR by heating the microcapillary while pulling it, thereby tapering the capillary³³ before fabricating the MBR. Although this method does not require the use of acids, the inner diameter is reduced when tapering the capillary - leading to problems with fluidic handling.

Biosensing with MBRs requires the accurate delivery of various solutions, such as silanes, specific antibodies, proteins, and other targets of interest; therefore, reliable fluidic handling is very important. One of the main advantages of using an MBR is the integrated fluidic channel provided by the capillary, which allows for efficient and targeted delivery of an analyte in the liquid or gaseous phase. This represents an improvement over other WGM microresonators that require additional external microfluidic channels for targeted analyte delivery^5,34,35. One challenge to using the integrated capillary for liquid delivery is the fluidic resistance in the microcapillary. Equation 1 shows that the resistance, R, to fluid flow is inversely proportional to the fourth power of the radius, r.

     (1)

Where η is the fluid viscosity, and L is the length of the fluidic channel. We compared fluidic resistance for two different capillary sizes, 75 μm inner diameter (ID) and 250 μm ID. The 75 μm ID capillary size could only pull ~10% of the target volume of DI water at a relatively low flow rate (100 μL/min). The 250 μm ID capillary pulled 90-100% of the target volume of DI water at the same flow rate (100 μL/min).

     (2)

Equation 2 shows the viscous resistance force, F_viscous, is directly proportional to the fluid velocity, v_m. Here, η represents fluid viscosity, and L is the length of the cylindrical channel. This equation shows that the viscous resistance force increases with fluid flow. This trend is also observed experimentally. This confirms our hypothesis that increasing the capillary diameter will decrease fluidic resistance in the capillary and improve the overall fluidic handling of the device.

In summary, we have shown a reliable protocol for fabricating reproducible WGM MBRs with integrated fluidic channels using the PGP, including quality control metrics. This protocol is simple, repeatable, and cost-effective. It can be further expanded to other heating methods, such as optical fiber splicers and CO₂ lasers. Additional improvements might include using high-pressure systems with inert gases like N₂ to control the inner pressure of the capillary during the fabrication process. Furthermore, we introduced protocols to increase the MBRs' sensitivity to biomolecule binding and bulk refractive index changes through HF etching. Lastly, capillary size was studied in terms of fluidic resistance to flow. Our findings showed that by increasing the inner diameter of the capillary, a reliable flow can be established to facilitate the accurate delivery of sensing targets.

The authors have nothing to disclose.

This project was supported in part by R41AI152745. AJQ was funded by the T32 Cancer Biology Award (NIH CA009547) and K08EB033409.