1. Introduction

Whispering gallery mode (WGM) optical microcavities [1–3] with high quality factors (Q) have attracted more and more attention in sensing research [4–9]. In traditional microcavity sensors, the resonant spectrum shift is detected to estimate the change of the refractive index (RI) of the cavity or the environment, and the sensing resolution depends on the Q greatly. Typically, ultrahigh Q microspheres, coupled by the fiber taper with nearly ideal coupling efficiency [10–12], are popularly used. However, there are some limitations of the microsphere-taper system when promoting the sensing research to practical application. First, the effective coupling can be affected, and even be broken when loading the vibration on the taper or the microsphere. Second, WGMs are very sensitive to surroundings, which is an advantage for sensors but makes the structure invalid in high RI materials or non-uniform dielectrics. Third, the exposure of the traditional coupling system makes the Q-maintenance challenging greatly, because the water and the dust in the air could spoil the Q drastically [13]. Finally, high-resolution 3D translation stages are necessary in the traditional coupling system [10], however, the stages are expensive and bulky, limiting the mobility in applications.

In this paper, for the first time, we propose and realize a novel packaged microsphere-taper coupling structure (PMTCS) experimentally. We demonstrate that the PMTCS can avoid the problems addressed above. The PMTCS is stable without 3D translation stages, robust against the vibration and free to move. Moreover, as a protective layer, the package body isolates the microsphere-taper from the surroundings, and maintains the Q above 10⁶ for a long time. In addition, by using the PMTCS, we realize a high sensitive and robust fiber-integrated thermal sensor which is potential for the distributed sensor network.

2. Fabrication

In experiments, individual microspheres are fabricated with the diameter (D) ranging from 180μm to 650μm by melting the end of the optical fiber. WGMs are excited by a tunable laser (1550nm wavelength band, linewidth < 300kHz) through a fiber taper. And the gap between microsphere and taper is controlled by electromechanical 3D X-Y-Z stages with 20nm resolution, as shown in Fig. 1(a).

 

Fig. 1 (a)–(d) Illustration of the packaging process. (e) A micrograph of a typical PMTCS structure.

Download Full Size | PDF

To package the microsphere-taper coupling system, ultraviolet (UV) glue is used. The glue is fibre coating material with low RI (n_g ≈ 1.35), which is made from special silicone acrylate with the product specification KD-310. The packaging process is comprised of four steps, as shown in Fig. 1(a)–(d). First of all, optimal coupling in the air with desirable resonant dips (Fig. 2(a)) needs to be achieved, as shown in Fig. 1(a). And then the UV glue is coated on the microsphere-taper by using a glue spreading machine in a dropping manner, as shown in Fig. 1(b). Afterward, the package is solidify through 10 minutes exposure under a UV lamp, as shown in Fig. 1(c). In the last step, the microsphere stem, mounted on the 3D stages, is truncated by using a heat burning manner. The final PMTCS is independent of the 3D stages and can be moved freely, as shown in Fig. 1(d). The whole procedure is monitored by two microscopes, horizontally and vertically. It is worth noting that we need to re-adjust the coupling before the solidification, because the initial coupling is affected by the glue dropping which changes the environment and the surface tension around the microsphere as well as the taper. The resonant spectra fluctuate randomly at the first moment of the glue dropping, and gradually stabilize when the glue inosculates with the coupling structure after a few minutes. Particular attention has also to be paid to avoid the taper cracking, especially at its fragile taper waist.

 

Fig. 2 (a) and (b) are contrast resonant spectra for a microsphere (D = 421μm) before and after the packaging, respectively. (c) Tested and fitted Q versus microsphere D.

Download Full Size | PDF

Figure 1(e) shows a typical micrograph (side view) of a PMTCS (D = 340μm), in which the package body is asymmetric due to the gravity. As the backbone, the tapered fiber traverses across the package body to hold the PMTCS. At the package boundary (marked with blue triangles), there are two contact points which cause extra scattering loss (less than 20%). In the package experiments, the scattering loss can be reduced through using a shorter taper and encapsulating it completely. This is also an effective way to improve the robustness which is determined by the un-stretched single-model fiber after packaging while depending on the fragile taper for an unpackaged system. In addition, the evanescent decay length (d) of WGMs ( $d=\lambda /\left(2\pi \sqrt{n_s^2-n^2}\right)1001\left[3\right]$, n_s and n are the RI of the microsphere and the surroundings, respectively.) increases from 0.1516λ to 0.3008λ, which causes a larger field overlap between WGMs and the taper. This makes the critical coupling easier to fulfill for a thicker taper. Besides, the coating layer with the minimum thickness about 30μm (≫ d) ensures the complete isolation of WGMs from the outside, as shown in Fig. 1(e).

3. Test results and analysis

By comparing the spectra before and after the packaging, a red shift to the longer wavelength (S) has been observed, which can be estimated through $S=\frac{{\lambda }^2}{\pi D}\left(\frac{1}{\sqrt{n_s^2-n_g^2}}-\frac{1}{\sqrt{n_s^2-n_{\mathit{air}}^2}}\right)$ [3], where n_air is the air RI. For a microsphere with D ≈ 421μm, the S we observed is 2.35nm, which agrees well with theoretical prediction of 2.3nm, as shown in Fig. 2(a) and Fig. 2(b). The package can also eliminate high radial order WGMs and offer much more regular spectra. Because after the package, the relative RI (n_s/n) decreases from 1.44 to 1.07, corresponding to an increase of the total reflection critical angle from 43.62 to 70.92 degree, which results in much larger leakage for high radial order WGMs [2].

As the most important parameter of the microsphere, the intrinsic Q (Q_tot) of WGMs can be expressed as

However, compared with the unpackaged structure, Q_tot decreases apparently in the PMTCS. As shown by red circles in Fig.2(c), the measured Q_tot decreases sharply when the diameter is less than 200μm, and Q_tot is always smaller than 10⁷ for larger microspheres. The similar phenomenon has been reported in microtoroids embedded in the water by Armani et al [15]. Former experiments have indicated that a coating on the microcavity surface can greatly reduce the Q_sca [16]. Thus, the loss in the PMTCS mainly originates from the radiation loss and the glue absorption. We fit the Q_tot (blue line) with n_g = 1.351 + 5 × 10⁻⁶i by analytically solving the WGMs in microsphere at λ = 1550nm (the formulas can be found in Ref. [12] and Ref. [13]). The results agree with the measurements greatly. Here, the imaginary part of n_g is corresponding to the amount of the absorption loss when the light propagates in the glue. With a real n_g=1.351, we obtain the Q_rad (green line) which decreases exponentially with the reduction in the diameter. In addition, particular attention should be paid to prevent contaminating the UV glue, because the contaminants attached to the microcavity or the taper can increase the overall scattering loss. On the contrary, mixing nano-particles at a certain concentration in the glue provides a feasible way to control the backscattering [17,18]. It is worth noting that the Q of the PMTCS is still much higher than the highest Q (2 × 10⁵) of a packaged microfiber coil resonator [19].

It is obvious that the package body isolates the whole coupling system from the surroundings, excluding Q spoiling factors from the dust and water in the air. As shown in Fig. 3(a), when exposing a microsphere in the air, the Q shows a quick decay due to the water absorption in a few minutes after its fabrication [13]. When putting it in the smoke, the Q has another remarkable decrease due to the scattering by the dust adhering to the surface. By contrast, the Q of PMTCS is much more stable and independent of the surrounding influences. In fact, we have maintained the Q above 10⁶ for a few months. Thus, the package provides a feasible way to maintain the Q, paving the way for devices research in practical application.

 

Fig. 3 (a) Tested Q versus elapsed time of a microsphere coupling system in the air and in the smoke, for un-packaged (in black) and packaged (in red) samples, respectively. (b) WGM wavelength shifts as a function of the surrounding temperature, in NaCl resolution and in pure water, respectively.

Download Full Size | PDF

4. Thermal sensing experiments

To verify the practicability of the PTMCS, thermal sensing experiments are carried out in complex dynamic water environment. A beaker containing 600ml water with a stirrer in it is adopted as the testing environment. The stirrer, not only helps to equalize the temperature, but also simulates a flowing water environment. A thermocouple and a heater are placed in the vicinity of the PMTCS, to measure and change the temperature, respectively.

The robustness is confirmed by the undisturbed spectra in the water flow. Furthermore, the completeness of the encapsulation is verified through the unshifted spectra when adding Sodium Chloride (NaCl) in the water gradually (keep the same temperature) to change the water RI about 2 × 10⁻³, which could cause a WGM wavelength shift about 20pm for an unpackaged silica microsphere [20]. Wavelength shifts against the temperature are recorded, in saturated NaCl solution and in pure water, respectively. As shown in Fig. 3(b), wavelength shifts are not affected by RI changes of the water. The temperature variation leads to changes both in the size and the RI of the silica microsphere, which subsequently causes resonance wavelength shifts [21–24]. The resonant wavelength shows a red shift about 160.39 pm when the temperature increases from 14°C to 26°C, which indicates a sensitivity of 13.37pm/°C with a relevant coefficient in linear fitting about 0.998. Taking into account the spectral resolution (Δλ_min) of our system, which is 0.015 pm, we estimate the resolution (ΔT_min = Δλ_min/(dλ/dT)) of the PMTCS microsphere temperature sensor as 1.1 × 10⁻³°C.

This resolution is in the same order of magnitude with the traditional silica microcavity thermal sensor [23]. What’s more, the practicability is greatly enhanced in the PMTCS, in which only the temperature change causes the WGM shift, while the change of external RI fails to cause the shift. Besides, the PMTCS thermal sensor shows an excellent anti-jamming ability, and can be used in harsh environments. Furthermore, the PMTCS can provide much more stable performance (for instance, the resolution) due to its superior Q maintenance ability. Additionally, the bulky translation stages are no longer unnecessary in the PMTCS. The portable structure makes these sensors easy to move and miniature, which is important in practical applications, especially in microsystem technology research.

Further efforts will be focused on exploring new package materials and improving the package technology to enhance the packaged Q_tot. The PMTCS could be extended to multi-packaged microsphere-taper system on one fiber to monitor the real-time temperature at different locations, or package multi-microspheres in a certain point to study the classical analog to electromagnetic induced transparency (EIT), even its application on gyroscopic [25, 26].

5. Conclusion

In summary, we have packaged silica microsphere-taper coupling systems by using the low RI UV glue. In our experiments, the (Q) of the packaged structure reaches 5 × 10⁶ which is limited by the radiation loss and the glue absorption. The anti-jamming ability and the isolation performance in PMTCS are verified experimentally. Temperature sensing experiments are also carried out. The results show that the PMTCS possesses a high resolution of 1.1 × 10⁻³°C with the remarkable practicability, robustness and convenience. Not only the PMTCS can be applied to microcavity thermal sensors, but promote the developments of other microcavity-based devices, such as the low threshold laser.